The model we have used to formalise the relationship between financial subsidies and the productivity of public enterprises draws mainly on work in the theory of incentives and the management of public enterprises, in particular the work of Holmström (1979) and Besley and Ghatak (2003) . These models emphasise the importance of incentive mechanisms in aligning the objectives of public enterprise managers with those of the state, in order to improve their performance and productivity.

We have incorporated the idea that financial subsidies can play a key role in providing resources for public enterprises to invest in infrastructure, build employee capacity and improve efficiency. In addition, our model takes into account the impact of external variables such as political stability, infrastructure investment, and economic inequality, following the principles of the public sector production model ( Ménard & Klein, 2004 ).

In this sense, our model seeks to analyse in a holistic way how subsidies influence the productivity of public enterprises, taking into account the structural and contextual factors specific to each country. The objective of this analysis is to assess the effects of financial subsidies on the productivity of public enterprises in Sub-Saharan African countries. In this respect, the productivity of public enterprises is chosen as the dependent variable. As a variable of interest, financial subsidies are considered as a determining factor.

These subsidies can be measured by the financial aid, direct or indirect, allocated by a public enterprise to finance an activity of general interest. The integration of these variables makes it possible to study their direct impact on the efficiency and performance of public enterprises, in a constantly changing economic context in Sub-Saharan Africa.

The functional form of the model may vary depending on the nature of the assumed relationship between financial subsidies and the productivity of public enterprises. A simple linear specification might be as follows:

${\text{PEPUB}}_{it}={\beta }_0+{\beta }_1{\text{FGRA}}_{it}+{\epsilon }_{it}$

A more complex specification of the model could include interactions between financial subsidies and other relevant variables. Firstly, the variable takes into account the context and structural characteristics of countries, such as political stability (POSTA), making it possible to measure the impact of political stability on the effectiveness of financial subsidies allocated to public companies.

Secondly, variables reflecting economic and social aspects, such as physical investment (MAIN) and education (EDUC), can modulate the impact of financial subsidies on the productivity of public enterprises. Thirdly, a variable representing income inequality (INCI), measured by the Gini index, could be used to assess how the distribution of wealth influences the effectiveness of subsidies and productivity.

Fourth, control variables such as trade openness (COPEN) and public health (RHEA), which respectively measure the impact of trade policies and health conditions on the productivity of public enterprises, taking into account the wider economic environment. These variables will make it possible to test interactive effects and better understand the underlying mechanisms linking financial subsidies to productivity in a diversified context.

Fifth, interactive variables, such as the interaction between financial subsidies (FGRA) and public health (RHEA), will be included. These interactions will make it possible to examine the extent to which financial subsidies can influence the productivity of public enterprises as a function of public health, and whether these effects are amplified or attenuated by other economic and social factors.

Sixthly, the measure of financial subsidies (FGRA) is the main explanatory variable in this analysis, as it is at the heart of the study. This variable can be measured by the amounts of direct or indirect financial aid allocated by the state or other public institutions to finance general interest activities within public companies. This variable of interest makes it possible to study the direct impact of financial subsidies on the productivity of public companies and to examine whether these effects are modified by variables specific to each country or sector.

With regard to the structural characteristics of countries, several factors can influence the effects of financial subsidies on the productivity of public enterprises. Among these, the level of economic development, physical infrastructure, national economic policies and political stability play a key role. These characteristics can modulate the way in which financial subsidies affect the productivity of public enterprises, for example, a stable and well-developed economic environment is more conducive to better use of subsidies to increase productivity.

Thus, studying the effects of financial subsidies on the productivity of public enterprises provides a better understanding of how public financing policies can improve the efficiency of public enterprises, taking into account their economic, political and social specificities. This analysis is essential to guide public policies aimed at optimising the use of financial subsidies to boost productivity, while taking into account the specific challenges of each country in Sub-Saharan Africa.

The functional form of the model is carefully chosen to capture the potential complexities of the relationship between financial subsidies and the productivity of public enterprises. Two main specifications are proposed to test the hypothesis that financial subsidies positively influence the productivity of public enterprises in Sub-Saharan Africa (SSA).

The first specification examines the direct impact of financial subsidies on the productivity of public enterprises, taking into account control variables such as the level of economic development, political stability and physical investment. This specification makes it possible to analyse how subsidies, as the main variable, directly affect the performance of public enterprises in different economic contexts.

The second specification includes interactive variables to assess the moderating effect of certain factors. For example, the interaction between financial subsidies and public health could reveal how a favourable health environment amplifies or attenuates the impact of financial subsidies on productivity. These two specifications aim to test the validity of the main hypothesis, while offering a more nuanced understanding of the impact of financial subsidies on the productivity of public enterprises in Sub-Saharan Africa, taking into account local specificities and contextual factors.

Direct relationship:

$\begin{array}{c}{\text{PEPUB}}_{it}={\beta }_0+{\beta }_1{\text{PEPU}}_{it-1}+{\beta }_2{\text{FGRA}}_{it}+{\beta }_3{\text{MAIN}}_{ij}+{\beta }_4{\text{EDUC}}_{ij}+{\beta }_5{\text{POSTA}}_{ij}\\ +{\beta }_6{\text{COPEN}}_{ij}+{\beta }_7{\text{PHEA}}_{it}+{\beta }_8{\text{INCI}}_{ij}+{\epsilon }_{it}\end{array}$

This specification focuses on the direct impact of financial subsidies on the productivity of public enterprises while controlling for other relevant factors. The coefficient β₁ represents the direct effect of financial subsidies on the productivity of public enterprises.

Indirect relationship:

$\begin{array}{c}{\text{PEPUB}}_{it}={\beta }_0+{\beta }_1{\text{PEPUB}}_{it-1}+{\beta }_2{\text{FGRA}}_{it}+{\beta }_3{\text{MAIN}}_{ij}+{\beta }_4{\text{EDUC}}_{ij}+{\beta }_5{\text{POSTA}}_{ij}\\ +{\beta }_6{\text{COPEN}}_{ij}+{\beta }_7{\text{PHEA}}_{it}+{\beta }_8{\text{INCI}}_{ij}+{\beta }_9\left(\text{FGRA}\ast \text{PHEA}\right)+{\epsilon }_{it}\end{array}$

This specification extends the direct relationship by incorporating interaction terms between financial subsidies and other variables:

This interaction term makes it possible to explore indirect effects, where financial subsidies influence the productivity of public companies through their impact on other factors. The coefficient ${\beta }_8$ represents this indirect effect ( Table 1 ).

Source: Authors.

These references are based on works and authors who have examined these variables in their research and are commonly used in studies of socio-economic effects and public policy. If you want more specific sources or recent articles, you can search for recent publications in academic journals.

The productivity of public enterprises (PEPUB) is a central concept that reflects the efficiency with which these entities utilise inputs “primarily labour and capital” to deliver goods or services in alignment with public interest goals. Following Baumol and Oates (1988) , the multiplicity of objectives pursued by public enterprises “ranging from economic to social missions” can complicate both the measurement and the optimization of their performance.

To ensure a laborious and operational measurement of PEPUB in this study, Total Factor Productivity (TFP) is adopted as the main indicator. In line with Harrison (1994) and Sheshinski and López-Calva (2003) , TFP captures the portion of output not explained by the amount of inputs used in production, thereby serving as a proxy for efficiency gains from better resource use or technological progress.

TFP is computed using an aggregate Cobb-Douglas production function of the form:

$Y_{it}=A_{it}{K}_{it}^{\alpha }{L}_{it}^{1-\alpha }$

where:

$Y_{it}=Y_{it}/K_{it}^{\alpha }{L}_{it}^{1-\alpha }$

The data for $Y_{it}$, $K_{it}$, and $L_{it}$ are extracted from the World Bank’s World Development Indicators (WDI) and IMF Government Finance Statistics (GFS), where available. The TFP index is standardized across countries and years to allow for meaningful comparison.

Regarding the key explanatory variable, financial subsidies to public enterprises (FGRA), it is scaled as a percentage of GDP in order to standardize across economies of different sizes. Thus, a one-unit change in FGRA corresponds to a 1 percentage point increase in the share of GDP allocated to subsidies for public enterprises. This scaling enables a clear interpretation of the marginal effect of subsidies on productivity, independent of country size ( Figure 1 ).

The evolution of the productivity of public enterprises (PEPUB) in Sub-Saharan Africa shows an overall positive trend over the period from 2000 to 2018, with a slow progressive increase. This increase reflects continuous but moderate growth. It suggests efforts to improve operational efficiency and resource management within public enterprises in the region, although minor fluctuations are observed from year to year.

Subsidies are defined as monetary transfers or cost reductions granted by the government to correct market failures and promote economic activities considered beneficial to society, according to Tirole (2014b) . He specifies that these aids can take the form of direct payments, tax reductions or other forms of financial support ( Figure 2 ).

Based on the WDI database, this figure shows a steady increase in FGRA in Sub-Saharan Africa over the period 2000 to 2018. Initially aligned with the productivity of public enterprises (PEPUB) in 2000, subsidies increased steadily over this period. This trend suggests a growing reliance on financial subsidies to support SOE operations, which could indicate persistent challenges in terms of financial self-sufficiency and efficient resource management. The steady increase in subsidies could also reflect increased needs for public funding to maintain the operation and growth of the sectors concerned.

The data used to estimate the model comes from various secondary sources. The data used in the study are panel data comprising a total of 43 countries over the period 2000-2018, selected on the basis of data availability. Most of the data used in this study comes from the World Development Indicators (WDI), the World Bank (WB) and the World Bank Enterprise Survey (WBES). The sample covers Sub-Saharan African countries. The inclusion of these 43 countries in the same sample can be justified by their strong historical and cultural roots, in addition to the economic links of being part of economic integration zones.

Table 2 shows the expected signs of the explanatory variables, based on existing theories.

Source: Authors.

The specification adopted in this study highlights two potential sources of endogeneity that can bias the results if they are not properly taken into account.

The first source of endogeneity arises from the inclusion of the lagged variable, thus introducing a time dependence into the relationship between the productivity of public enterprises (PEPUB) and financial subsidies. The inclusion of the lagged productivity variable in the model can lead to a correlation between the dependent variable and the specification error, as the effect of past productivity can influence both financial subsidies and current productivity.

The second source of endogeneity lies in the inverse causality between the productivity of public enterprises (PEPUB) and financial subsidies. It is indeed plausible that an increase in the productivity of public enterprises leads to an increase in financial subsidies, or that these subsidies have an effect on current productivity. This bidirectional relationship complicates the estimation of causal effects, as it introduces a bias into the relationships between the variables in the model.

To deal with these endogeneity problems, the Generalised Method of Moments (GMM) proves to be the most appropriate technique. This method, developed by Arellano and Bond (1991) , is based on the use of instruments to overcome endogeneity problems by resorting to exogenous or instrumental variables. This approach has been widely used in similar empirical work, particularly in studies that focus on dynamic panel data models where the explanatory variables may be correlated with the errors in the model.

The first estimator used in the MMG method is the Diff-MMG estimator proposed by Arellano and Bond (1991) , which is based on first differences of the dependent variable and the use of instruments from the lagged variables. This estimator is particularly suited to situations where the number of periods (T) is relatively high compared with the number of units (N). However, in the case of this study, where T = 19 and N = 43, this estimator may have limitations. Indeed, when the number of periods is low, it becomes necessary to use a more robust version of the estimator.

The Sys-MMG estimator, developed by Blundell and Bond (1998) , is an important improvement on the Diff-MMG method. It combines both first differences and variable levels to provide more robust and less biased estimates, especially in cases where the number of periods is small and the explanatory variables are not strictly exogenous. This approach is particularly suitable in contexts where conventional instrumentation may not be powerful enough to solve endogeneity problems ( Blundell & Bond, 1998 ).

In addition, this estimator has been widely used in empirical studies dealing with the impact of economic policies on economic performance in panel data, as in the work of Roodman (2009) , who proposed an improvement to the estimation methodology using the MMG method to resolve endogeneity problems.

Building on this previous work, this study applies the MMG method to address endogeneity problems due to the lagged variable and reverse causality between public firm productivity and financial subsidies. For example, the work of Arellano and Bond (1991) showed that this technique is particularly effective in dealing with endogeneity in dynamic panel data models, and Blundell and Bond (1998) demonstrated that the Sys-MMG estimator improves estimates when the explanatory variables are correlated with the model errors.

The Diff-GMM and Sys-GMM estimators are therefore applied in this study to produce reliable, robust and consistent estimates, overcoming endogeneity problems and dealing with reverse causality between financial subsidies and the productivity of public enterprises. This methodological choice is supported by a large literature that has shown the effectiveness of this approach in similar contexts, notably in the work of Arellano and Bond (1991) , Blundell and Bond (1998) and Roodman (2009) , all of whom have used the MMG method to address endogeneity issues in dynamic economic models.

According to Table 3 , PEPUB over the study period has a mean of 5.292; a standard deviation of 0.664; with a minimum and maximum of 3.487 and 7.098, respectively. Financial subsidies in SSA have an average of 0.324 with a standard deviation of 0.095; a minimum value of 0.051 and a maximum value of 0.568 for the period under consideration. Trade openness has an average of 0.201; standard deviation of 0.401, with a minimum of 0.121 and a maximum of 0.952. Education has an average of 0.420 and a standard deviation of 0.130, with a minimum and maximum value of 0.011 and 0.923.

Source: Auteurs.

Table 4 shows the correlation between the various main variables. The correlation matrix is used as a statistical measure that expresses the notion of a linear relationship between variables (meaning that they move together at a constant rate). It is a common tool for describing simple relationships without worrying about cause and effect. In the majority of cases, the cross-correlation is significant, positive, but less than 0.50, which removes the threat of multi-collinearity between the control variables.

Source: Authors.

This section presents the results, their interpretation, as well as discussions and policy recommendations. Robustness analysis will also be carried out. Table 5 presents the results of the basic regression model using the full sample specification. This specification includes the two-stage MMG-System estimator incorporating Windmeijer’s (2005) finite sample standard error correction method. In addition, other consistent but potentially less efficient estimators are sent in advance. These estimators all deal with econometric problems induced by unobserved country-specific effects and potential endogeneity.

Source: Authors. Note: Column (1) (MMG in system; direct relationship) considers the maximum of the predetermined variables as instruments of order 2. Column (2) (MMG in system; indirect relationship) presents the result of the two-stage interactive variable. Column (3) presents the result of the marginal effects obtained from the direct relationship. The fixed effects by period and by country are used but not reported in this table. The estimates were carried out in the second stage. The figures in brackets correspond to the robust standard deviations. (*, **, ***) represent significance at the 10%, 5% and 1% thresholds respectively (p < 0.1, p < 0.05, p < 0.01).

In Columns (1), (2) and (3), there is a positive relationship between financial subsidies and the productivity of public enterprises. The results of the method of generalized moments in system (direct relationship, indirect relationship and marginal effect) will be interpreted. The lagged dependent variable is significant and its magnitude reveals that the productivity levels of public enterprises change only slowly over time and depend on past levels. This result also justifies the specification of the dynamic model and the use of the MMG approach in system, because of the problem of the weakness of the instruments associated with the MMG in difference when the series is very persistent.

Firstly, by taking into account the delayed PEPUB, it is possible to calculate the direct long-term effect of FGRA on PEPUB. The estimates given in Column (1) of the table imply that the long-run effect of a one-unit change in FGRA is to increase PEPUB by about 13.353 (=0.681/(1 − 0.949). The coefficient on the lagged productivity variable for public enterprises (PEPUB_t−1) is highly significant (0.949) with a standard deviation of 0.014, which means that the productivity of public enterprises is highly autocorrelated over time.

In other words, past productivity strongly influences the current productivity of state-owned enterprises in Sub-Saharan Africa. This result is consistent with that of Balisacan et al. (2011) , who found a positive and significant relationship between the productivity of SOEs over time. However, our study adds novelty by highlighting that this relationship is particularly strong in the context of developing countries, showing the high dependence of SOEs on past performance to maintain their efficiency.

Second, the validity of the interactive effects between FGRA and public health was tested. In Column (2), the analysis of the interactive effects between financial subsidies (FGRA) and public health shows a coefficient of 0.402 for the interactive variable, indicating a positive and significant effect on the productivity of public enterprises (PEPUB). In concrete terms, a one-point increase in the interactive variable between FGRA and public health leads to a 40.20% increase in PEPUB.

This result suggests that financial subsidies and public health are complementary. In other words, financial subsidies improve the productivity of public enterprises only by strengthening public health. This result is in line with the work of Smith and Haddad (2002) , who highlighted the impact of subsidies on the productivity of public enterprises through health-related channels. Nevertheless, our study offers a unique contribution by demonstrating that this effect is particularly marked in the context of Sub-Saharan Africa, where public health is a key factor in the performance of public enterprises.

Thirdly, the marginal effects were calculated in order to see the effect of each of the explanatory and interactive variables independently of the others (ceteris paribus). We observe that the increase of one unit of financial subsidy (FGRA) leads to a direct increase of 67.70% in the productivity of public enterprises, all other things being equal. This result highlights the direct impact of financial subsidies on the productivity of public enterprises in Sub-Saharan Africa.

However, when we consider the interactive effects, it appears that the emphasis must be placed on improving public health in order to reinforce the effects of subsidies on productivity. This finding is in line with the idea that healthy individuals are more likely to work and produce more than sick individuals. This point is in line with the research of Bloom and Canning (2000) , who demonstrated that health has a direct impact on productivity. In the case of Sub-Saharan Africa, integrating public health policies into financial subsidy strategies could therefore maximise the impact on the productivity of public enterprises.

As the results show, the coefficient on financial subsidies (FGRA) is positive and highly significant (0.681), suggesting that an increase in the financial subsidies allocated to public enterprises has a positive effect on their productivity. In other words, as financial subsidies increase, SOEs in Sub-Saharan Africa become more productive.

This finding is consistent with the work of Tarp (2013) , who has shown that subsidies have a direct impact on firm productivity in developing countries. The difference here is that our study shows the magnitude of this impact in a specific regional setting, Sub-Saharan Africa, where subsidies play a crucial role in supporting state-owned enterprises, compared to other regions that are less dependent on public subsidies.

The coefficient on physical investment (MAIN) is also positive and significant (0.921), indicating that the increase in physical investment in public companies leads to a substantial increase in their productivity. This result supports the idea that investment in infrastructure and equipment is essential for improving the productive capacity of companies.

A similar result was found by Krugman and Obstfeld (2009) , who emphasise that physical investment is a key driver of economic growth. However, our study highlights that this effect is even more pronounced in the context of Sub-Saharan Africa, where infrastructure is often inadequate and investment is crucial to economic development.

The empirical results also confirm that education and public health can be important determinants of the productivity of public enterprises. The coefficient on education (EDUC) is positive and significant (0.770), which shows that an increase in the secondary school enrolment rate leads to better productivity in public enterprises. This reflects the importance of human capital in improving the performance of public enterprises.

This result is consistent with the work of Barro (2001) , who demonstrated that human capital is an essential determinant of economic growth. However, in our study, the impact of education is more marked in countries in the Sub-Saharan African region, where improving education can have a greater direct effect on the productivity of state-owned enterprises, which are often characterised by a less skilled workforce.

However, the coefficient on political stability (POSTA) is positive and significant at the 5% level (0.041), suggesting that greater political stability is associated with higher SOE productivity. Greater political stability promotes a more predictable economic environment, which is essential for the efficiency of public enterprises.

This result is in line with Acemoglu and Robinson (2012) , who have shown that political stability is a key factor for economic growth. Nevertheless, our study makes a significant contribution by highlighting that, in the context of Sub-Saharan Africa, political stability has a particularly important impact on state-owned enterprises, which are often affected by frequent political instabilities.

Income inequality has a negative and significant effect on the productivity of public companies. The coefficient on income inequality (INCI) is negative and significant (−0.014), suggesting that higher income inequality has a negative effect on the productivity of public enterprises. This reflects the idea that excessive inequality can generate social tensions and reduce the overall efficiency of public enterprises by lowering the general well-being of the population.

This result is consistent with the work of Deininger and Squire (1996) , who showed that income inequality has negative effects on economic growth. However, our study stands out by highlighting the impact of inequality on the productivity of state-owned companies in Sub-Saharan Africa, a region where inequality is particularly high and can exacerbate social and political tensions.

The coefficient of trade openness (COPEN) is positive and highly significant (0.018), indicating that trade openness has a positive effect on the productivity of public enterprises. This suggests that trade liberalisation and integration into world markets contribute to increasing the efficiency of public enterprises by exposing them to increased competition and new export opportunities. This result is consistent with the work of Frankel and Romer (1999) , who found a positive link between trade openness and economic growth. Our contribution lies in the specific application to Sub-Saharan Africa, where trade openness has more visible effects due to growing regional integration and initiatives such as the African Continental Free Trade Area (AfCFTA).

The coefficient on public health (RHEA) is positive and significant at the 5% level (0.004), indicating that improving public health has a positive impact on the productivity of public companies. A healthier population leads to a more productive and fitter workforce, which promotes the growth of public companies. This result is in line with the work of Bloom and Canning (2000) , who demonstrated the link between health and economic productivity. However, our study differs in its focus on Sub-Saharan Africa, where health challenges are particularly pronounced and where improvements in this area can have a more direct and visible impact on the productivity of public enterprises.

In addition, as can be seen in the lower part of the table, the model specification is supported by all available specification tests, so that misspecification or poor instrumentation do not appear to be driving the results. The first-order autocorrelation, AR(1), is present in the data, as expected. Second-order autocorrelation AR(2) is absent from the data, which must be the case for the MMG estimator to be consistent.

The Hansen/Sargan test does not reject the over-identifying restrictions (instrument exogeneity) at conventional significance levels, indicating that the instruments are valid. The Hansen/Sargan test suggests that the (additional) MMG-System hypothesis is not violated and that the additional instruments introduced are valid, which further strengthens the statistical properties of the results.

In Column (1) of the table, two steps are taken to limit the number of instruments. Firstly, the lag ranges are limited and only three appropriate lags of each endogenous explanatory variable are used in the difference equation and the first contemporaneous difference as an instrument in the levels equation. Secondly, the set of instruments has been “collapsed”, which is equivalent to combining the columns of the instrument matrix by addition.

The estimation procedure then uses a common variance-covariance of the moment conditions across periods ( Roodman, 2009 ). At the cost of reduced efficiency, these two steps can significantly reduce the problems caused by instrument proliferation by accounting for cases where the unrestricted variance-covariance is too large for estimation and inversion ( Loayza et al., 2012 ). This can be particularly crucial in the case of a large number of explanatory variables and the presence of multiple time periods.

In order to make a comparison and assess the robustness of the results according to different model specifications, the results are presented according to level of economic development and region. The study distinguishes between Least Developed Countries (LDCs) and Developing Countries (DCs). This distinction is based on the work of Churchill and Yew (2017) , who suggest that the effects of financial subsidies may differ between countries depending on their level of development.

To do this, the study adopts the United Nations classification, as defined in the 2024 World Economic Situation and Prospects report. The advantage of this system over that of the World Bank is that it is not limited solely to national income. It also takes into account various other criteria to determine whether a country can be considered developed. This approach enables a more detailed analysis of the disparities between countries and a better identification of the factors that have a variable influence on their food security ( Table 6 , Table 7 ).

Source: Authors. Note: Figures in brackets indicate robust standard errors. (*, **, ***) represent significance at the 10%, 5% and 1% thresholds, respectively (p < 0.1, p < 0.05, p < 0.01). Column (1) presents the direct relationship between the financial subsidy and the productivity of public enterprises in SSA. Column (2) presents the transmission channel test.

Source: Authors. Note: Figures in brackets indicate robust standard errors. (*, **, ***) represent significance at the 10%, 5% and 1% thresholds, respectively (p < 0.1, p < 0.05, p < 0.01). Column (1) presents the direct relationship between the financial subsidy and the productivity of public enterprises in SSA. Column (2) presents the transmission channel test.

In order to compare the effect of the FGRA on the PEPUB by level of economic development of the countries, the sample is made up first of the least developed countries and then of the developing countries in SSA. The “MMG-type” instruments have been replaced by their main components. The results remain qualitatively the same in LDCs and DCs. The coefficients of the “FGRA” variable remain positive and significantly different from zero in both cases ( Table 6 , Table 7 ).

However, it should be noted that the FGRA has a greater effect (direct and indirect) in the least developed countries than in the developing countries. It emerges that a one percent increase in FGRA leads to a direct increase in PEPUB of 85.40% in the LDCs compared with an increase of 70.30% in the DCs. The need for FGRA is therefore higher in LDCs. This can be explained by the fact that developing countries have the best public health indices. As mentioned above, a healthy person is more willing to work than a sick person. A FGRA access policy based on people’s health should be favoured more in the LDCs than in the DCs in order to catch up with the DCs.

The remainder of this paper carries out a number of additional analyses to assess the sensitivity of the results to regional characteristics. To do this, the model will be re-estimated by grouping countries according to SSA economic communities. As the conditions for applying the MMG as a system are not met, the models will be re-estimated using the double least squares method and the POOLED OLS method to resolve the potential endogeneity problem. Referring to the World Bank classification, we group the sample countries into five economic communities, namely ECCAS, ECOWAS, COMESA, EAC and SADC ( Table 8 ).

Source: Authors. Note: Figures in brackets indicate solid standard errors. (*, **, ***) represent significance at the 10%, 5% and 1% thresholds, respectively (p < 0.1, p < 0.05, p < 0.01).

The sub-regional results reveal that financial subsidies are positively associated with the productivity of public enterprises across the various economic communities in Sub-Saharan Africa (SSA). An increase in financial subsidies leads to a statistically significant rise in public enterprise productivity, and this pattern remains qualitatively consistent with the results obtained for SSA as a whole. This overall trend may be partly explained by the widespread impact of recent health and security crises, which have compelled many governments to strengthen their financial support for state-owned enterprises.

However, the magnitude of this effect varies across regions, reflecting underlying economic, institutional, and structural differences. The results indicate that the impact of financial subsidies is most pronounced in Eastern and Southern Africa, specifically within COMESA (Common Market for Eastern and Southern Africa) and SADC (Southern African Development Community). In these regions, a 1 percentage point increase in financial subsidies as a share of GDP is associated with a remarkable 94.10% and 93.80% increase in the productivity of public enterprises, respectively.

By contrast, in Central Africa (ECCAS) and West Africa (ECOWAS), the effects, while still positive and significant, are somewhat less substantial. A 1 percentage point rise in subsidies corresponds to an 89.60% and 89.90% increase in productivity in these regions, respectively. The lowest effect is observed in East Africa (outside SADC), where the same increase in financial support leads to a 75.10% rise in productivity.

These regional disparities may be attributed to several contextual factors. Regions such as COMESA and SADC may benefit from more effective institutional frameworks, greater administrative capacity, and stronger policy coordination mechanisms, which enhance the efficient use of financial resources. Furthermore, these regions have made notable progress in public sector reform and infrastructure development, potentially enabling public enterprises to better leverage financial subsidies. Conversely, ECCAS and ECOWAS face enduring challenges such as institutional fragility, governance deficits, and underperforming state-owned enterprises, which may limit the effectiveness of subsidies in stimulating productivity.

Importantly, all control variables retain their expected signs and statistical significance, reinforcing the robustness of the empirical findings. Overall, the results suggest that financial subsidies are a crucial tool for enhancing the performance of public enterprises, but their impact is highly contingent on regional governance capacity and structural conditions.