The literature review reveals a fragmented consensus on the definition of the CE, despite research dating back to the 1960s [13]. While many scholars concur that CE encompasses the reduction, reuse, and recycling of raw materials (the 3Rs), others argue that CE transcends environmental considerations to incorporate socio-economic strategies, thereby contributing to global well-being [14]. The challenge of defining CE's scope, akin to an "umbrella concept" [15], complicates the selection of indicators for measuring circularity [16]. To address the gaps in CE monitoring, some studies [17] suggest developing a composite index, though selecting inclusive criteria remains a challenge, necessitating a methodical approach. Studies indicate that the context and geographical implementation area can influence the sustainability and circularity pathways [18]. Indeed, the CE can stray from long-term environmental sustainability when product transformation processes shift one type of pollution to another [19]. A holistic approach ensures interdependence and prevents the focus from being limited to a single Sustainable Development Goal (SDG) [20]. The multidimensional nature of the CE raises questions about prioritising challenges in each context and how different dimensions interact. Addressing CE's main challenge, resource conservation in production and consumption, highlights the critical role of food and energy resources, particularly for African countries [21]. Food and energy are closely linked [22], with rising energy prices affecting the costs of food production, storage, transport, and distribution. Nexus thinking positions the CE as part of an integrated and interdependent sustainability strategy.

The literature review underscores the pivotal role of governments in fostering the institutional framework and implementation of a transformative change process conducive to the transition to a CE [23]. Furthermore, it identifies numerous barriers and drivers influencing the transition to a CE, with specific implications for the African context [24]. Technical barriers are predominant, as countries require not only access to circular technical solutions but also the capability to implement these solutions swiftly. Information and communication technologies play a crucial role in enhancing these technical capabilities [25]. Technical skills present a significant challenge in Africa, often misaligned with the continent's unique needs and conditions. For example, construction techniques in sub-Saharan Africa mimic those of developed countries, frequently ignoring the local climate and leading to increased energy consumption due to the lack of an adaptive bioclimatic design approach [26]. The excessive dependence on unsustainable techniques, such as the use of off-grid solar technology, results in substantial waste and short product lifespans in the region [27]. Additionally, the scarcity of repair skills and the prevalence of a large informal market undermine countries' technical capabilities [28]. Economic barriers also play a role, with market uncertainties and high costs deterring investment.

Furthermore, socio-economic inertia serves to exacerbate the reliance on unsustainable solutions and to restrict funding for innovation, especially among small and medium-sized enterprises (SMEs) [29]. A number of studies have highlighted that African countries often overexploit resources without satisfying their needs, a situation that is further compounded by low levels of production investment, inadequate infrastructure, suboptimal resource management, and a lack of human resource capacity [22]. Nevertheless, the vulnerability caused by resource depletion or price volatility can also drive the shift towards CE [23].

In order to identify the factors influencing the transition trajectories towards a circular and sustainable economy in African countries, the remainder of the paper is structured as follows: Section 2 outlines the method for assessing the impact of barriers and drivers on the CE trajectory in African countries, addressing the challenges of data collection. Section 3 presents a typology of African countries based on their CE efforts, models the factors influencing CE in Africa, and offers recommendations to enhance circularity. Section 4 concludes with the main findings of the research.

Methods for constructing a composite index [30] propose a multi-step approach: defining the object of study with reference to a theoretical context, selecting suitable simple indicators, standardising each indicator and choosing the aggregation method. While there is consensus on the above steps, it is crucial to recognise that the substitutability or non-substitutability of the indicators selected, the aggregation method (complex or simple), the relative or absolute comparison between countries and the indicator weighting method all have a significant impact on the construction of the CESDI and CESDIaag indices.

Normalisation is required before aggregation and allows for the comparison of indicators on different scales by transforming them into normalised values in the range 0-1. The normalisation method depends on the type of comparison (absolute or relative). A relative comparison was chosen to assess and benchmark the performance of different African countries effectively. The value of the normalised index that tends to 1 indicates a significant contribution to the country's circularity compared to other African countries. On the other hand, when the value reaches 0, the country's performance is weak compared to all African countries.

Table 3 summarises the normalisation approach by presenting selected indicators with different units, each reflecting a key dimension of the CE in Africa.

The data are collected for the same year to facilitate comparison. The contribution column specifies whether an indicator positively or negatively affects CE performance. The maxX and minX columns indicate the countries with the highest and lowest values for each indicator, respectively. These insights reveal regional disparities and diverse circular economy trajectories across Africa, facilitating cross-country comparisons and highlighting sustainability performance gaps.

The max-min method is applied for normalisation, utilising equations (1) and (2). For both equations, y j i.s represents the normalised value of the indicator I^i,s for the given country j, while Xji.sdenotes the raw value of the same indicator for country j.

The parameters are defined as follows: j varies from 1 to N where N = 54 represents the number of African countries, i ranges from 1 to 12 representing different indicators included in the construction of CESDI and CESDIaag indices, s takes the value 0 when the indicator is directly integrated into the calculation of CESDI and CESDIaag values and ranges from 1 to 3 for indicators used to construct the environmental regulation composite indicator y^6,0, maxX^i,s and minX^i,s are the maximum and minimum values of Xji.s across all countries.

Alternatively

Eq. (1) normalises data where the high value indicates more circularity and contribution to Africa's sustainable development challenges, such as agricultural land, total renewable water resources and other indicators mentioned in Table 3. Eq. (2), on the other hand, normalises data where a low value indicates more contribution to CE, such as waste generation, air pollution and domestic material consumption.

The following example illustrates the calculation method using eq. (1) for the normalised indicator of Agricultural land per capita in Gabon. In this case, i = 1, s = 0 and j = Gabon. The raw value of the indicator is XGabon1.0=0.0317. From Table 3, the raw maximum value is maxX ^1.0 = 0.16 (observed in Namibia), and the raw minimum value is minX ^1.0 = 0.00016 (observed in Seychelles).

The normalised indicator value is then calculated as follows:

Another example illustrates how eq. (2) can be used to calculate the normalised indicator y^5,0 for air pollution in Gabon. In this case, i = 5, s = 0. The raw air pollution value for Gabon is XGabon5.0=0.2474, recorded in the database [34]. From Table 3, the raw minimum value is minX^5.0= 0.0256, observed in the Central African Republic, while the raw maximum value, maxX^5.0= 8.480, is recorded in Uganda.

The normalised indicator value is then calculated as:

For CESDIaag, Xji.s is replaced by Xjaag(Tf−Tk)i.s in equations (1) and (2) to normalise the average annual growth for the rate of indicator I^i,s over a given period (Tf−Tk).Xjaag(Tf−Tk)i.Srepresents the average annual growth rate of the indicator Xji.s over the period (T_f‒T_k) as show in eq.(3),maxXaag i.s and minXaagi.s are the maximun and minimun values of Xjaag(Tf−Tk)i.s across all countries. For each indicator, T_f is the most recent date and T_k is the initial date. So,

Thus, in the case of Gabon, and referring to the indicator agricultural land per capita, T_f corresponds to the year 2016 and T_k to the year 2010, as specified in Table 2, so ⁽T_f ‒ T_k) = 6. On this basis, it can be deduced that:

The normalised value is then deduced by applying eq. (1), given that indicator values are max Xaag1.0=−0.00428 and minXaag1.0=−0.0839, so yaag(6)Gabon1.0(XGabon)=0.617.

The normalised data are aggregated. Thus, the CESDI is a composite of three dimensions having the same ponderation as mentioned in eq. (4).

Each dimension is composed of four equally weighted indicators, as explained in eq. (5).

Thus, the composite CESDI index is the sum of 12 indicators, as specified in eq. (6). All integrated indicators are simple, except for that related to environmental regulation, which is itself a composite indicator consisting of three indicators, as mentioned in eq. (7). So,

Where ∝ _i, is the weight given to the i-th indicator. The value of ∝ _i indicates the degree of importance of each variable in the construction of the index. The choice of the weight given to each dimension is an arbitrary decision [30]. In this research, the same weight (equal to 13×14 ) is assigned to the indicators to emphasise the equal importance of each dimension and each indicator, while the variables are combined using an additive function.

The classification of countries based on the method of nested averages will make it possible to establish a typology of African countries in terms of their static and dynamic circular economy performances.

In order to measure the impact of barriers and drivers on the CE trajectory of African countries, the following models represented by equations (8) and (9) are considered:

Where CESDI and CESDIaag are the dependent variables, βk(1)and βk(2)are the coefficients of the regressions and Z_{k j} are the different independent variables for the countries identified by subscript j, ε is the error term.

In this study, Stata software was used to perform Ordinary Least Squares (OLS) regression analysis to estimate the association between the independent variables (CESDI, CESDIaag) and different barriers and drivers affecting the CE. OLS regression, a linear modelling technique, was chosen for its ability to model relationships involving multiple dependent and independent variables [44]. The factors influencing CE (underlined names in Table 4) are indicative of the barriers and drivers discussed in the literature presented in the first section Some factors were omitted from the models due to multicollinearity concerns or other considerations.

The OLS method is a statistical technique that seeks to minimise the sum of the squared differences (residuals) between observed values and those predicted by the model. The estimated coefficients β^kquantify the individual contribution of each independent variable, indicating the marginal effect of each explanatory variable on the dependent variable. Under the standard assumptions of linearity, homoscedasticity, and error independence, OLS produces unbiased and efficient estimators with the lowest possible variance. Hypothesis testing to assess the statistical significance of an explanatory variable is performed on its corresponding β coefficient. The null hypothesis (Ho: β_k = 0) states that the variable has no effect on the dependent variable. The standard error (SE) quantifies the uncertainty surrounding the estimation of the coefficient β^k . It is directly involved in the calculation of the Student’s t-test statistic, presented in eq. (10), which is used to evaluate the validity of the null hypothesis. If this hypothesis is rejected, it suggests that the variable significantly contributes to predicting or explaining the target variable.

Figure 2 provides an analysis of the static and dynamic circular performances of African countries in terms of food (CEFS, CEFSaag), energy (CEEA, CEEAaag) and resource efficiency (CEER, CEERaag).

Figure 2.

Performance levels of African countries related to CEFS, CEFSaag, CEEA, CEEAaag, CEER and CEERaag

Regarding food, the CEFS map shows that neighbouring countries, both low- and high-CEFS ones, face similar water, land and deforestation nexus for countries. This finding is confirmed by the CEFSaag map, which delineates the dynamic perspective and reveals a clustering of low dynamic countries in the central and western regions facing significant challenges, in particular, political conflicts that impede the rapid implementation of essential economic and political reforms in the agricultural sector [48]. For energy, the CEEA map shows that the main producers of fossil energy (oil and gas) in Africa, such as Nigeria, Algeria, Libya and Egypt [49], are not well classified according to CEEA. The abundance of fossil natural resources prevents the consideration of long-term strategies, such as the transition to renewable energy and the reduction of pollution. Renewable energy is not considered as part of a sustainable and equitable energy access strategy for African countries, as also mentioned by other studies [50]. The CEEAaag map shows that Malawi, Liberia, Rwanda, Ethiopia and Seychelles have made the most efforts to strengthen energy availability. At the same time, Algeria, Chad, Djibouti and Senegal are the least developed countries on this axis.

In terms of resource efficiency, the CEER map highlights the ability of African countries to decouple growth from resource use. It indicates that South Africa, the Republic of the Congo, Equatorial Guinea, and Mauritius are the best performers in the static approach (CEER). Using the dynamic approach, Uganda, Comoros, Ethiopia, Sudan, Mauritius, and Eswatini (Swaziland) have made the most progress.

Appendix Tables A1 and A2 show the results of applying the CESDI and CESDIaag indices to 54 African countries and compare their performance in static and dynamic approaches. This ranking serves as the basis for defining a typology of African countries in terms of their circular and sustainable performances, as illustrated in Figure 3.

Figure 3.

Typology of African countries according to CESDI and CESDIaag performances

Each country is placed on the graph according to its performance on the CESDI (horizontal axis) and the CESDIaag (vertical axis). Thus, the first group, "constantly moving forward", is made up of countries whose CESDI and CESDIaag performances are above average. These countries are in an interesting CE dynamic. This group includes five island countries: Cape Verde, São Tome and Principe, Comoros, Mauritius and the Seychelles. Despite their vulnerability to climate change, they have risen to the challenge of implementing optimal resource management. Group 2 "be awake" is made up of countries that have a low CESDI but a high CESDIaag. Some disadvantaged and low-development countries, particularly those in the eastern region, are making good progress that will allow them to catch up and move to CE in the future. These include countries such as Ethiopia, Somalia, Rwanda, Lesotho and Burundi.

On the other hand, group 3 "stay stagnant" is made up of countries with high CESDI and low CESDIaag values. The countries in this group have interesting circular performances compared to other African countries. Still, they are not in a circular dynamic that could allow for the prediction of future progress. Many of these countries are in West Africa. Some of them have benefited from an abundance of natural resources, especially fossil resources. The last group, "unable", represents the countries that are in the most critical situation with a low CESDI and a low CESDIaag. It includes countries such as South Sudan, Niger, Libya and Togo.

The results of the regressions in Table 4 provide further explanation of the typology of African countries. The R-squared indicates that more than 97% of CESDI and CESDIaag are explained by the independent variables, which confirms the goodness of fit of the models. Additionally, the significance levels reflect the statistical confidence in the observed relationships: *** denotes a highly significant result with a probability of randomness below 1%, ** indicates statistical significance with a probability below 5%, and * represents moderate significance with a probability of randomness up to 10%.

Model 1 results demonstrate a positive relation between CESDI and variable HDI. This relationship is significant at a 99% level and indicates that human development has an impact on the ability of African countries to adopt circular and sustainable approaches. The higher a country's level of development, the more likely it is to adopt the principles of a CE. Conversely, low development represents a significant barrier to the transition to a CE. This result is in line with Beckerman's work [51] on the Kuznets Environmental Curve, which demonstrates that development is the key to better environmental quality. The positive and significant relationship between CESDIaag and HDI in Model 2 confirms the importance of human development as the driving force behind a circular trajectory.

The estimation results show a significant and negative relationship between resource rents (variable Rent) and CESDIaag, indicating that the more the economy of an African country is based on resource rents, the weaker the dynamics of the transition towards a circular economy.

The government effectiveness variable (Government) is significant and negative in Model 2. Circular dynamics are highest in countries where the perception of the quality of public services is lowest. This result may appear counterintuitive, but it underscores the intricate nexus between the quality of institutions and the pressure on resources in the African context. Indeed, some studies have demonstrated that low institutional quality is indicative of a high level of corruption, which reduces the attractiveness of resource-intensive projects, thereby contributing to enhanced environmental preservation [52]. Consequently, the quality of institutions does not appear to be a driver behind the transition to the CE in Africa.

This finding is corroborated in countries that also exhibit unsustainable infrastructure, as measured by the proportion of urban population living in slums (variable Slums). The positive and significant relationship between CESDIaag and the percentage of the population living in slums shows that it is in countries with the highest rate of slums that progress in resource conservation is most significant. Thus, the quality of the infrastructure is conducive to resource-intensive projects. The results of the estimated models show that the transfer of technical skills (variable Technic) and direct foreign investment (variable FDI) do not have a significant impact on the transition to CE in the African context. In this context, at present, the transfer of techniques does not contribute to resource circularity.