This study assesses the effects of COVID-19 control measures on the resilience of economic growth. We applied the ordinary least squares method on a sample of 11 Central African countries with daily data from 2020 to 2021. According to the results, measures to combat COVID-19 (income support, fiscal debt relief, closure of schools and workplaces, cancellation of public events, quarantine, handwashing and lockdown) decrease the ability of economic growth to withstand the COVID-19 shock and to return to equilibrium after the shock. Furthermore, the results of the mediation analysis show that the effects of COVID-19 control measures (income support and debt relief) on the resilience of economic growth is mediated by corruption. From a policy perspective, we suggest strengthening political, economic and health institutions to combat future shocks.
The HIV-19 pandemic has pushed Africa into its worst recession in more than 50 years, resulting in a 2.1% decline in gross domestic product in 2020 and plunging an estimated 30 million Africans into extreme poverty by 2021. According to the AFDB report (2021, 2022)1,2, the continent has lost about 22 million jobs by 2021. Similarly, the World Bank report (2020) reveals that global economic activity will contract to 2.8% in 2020 from 2.2% in 2019. In the same vein, a recession of −4.4% is expected, which is worse than that observed during the 2008 financial crisis.
COVID-19 has been identified by economic literature as having various determinants, including urbanization (Nguimkeu & Tadadjeu, 2021) , inequality (Okoi & Bwawa, 2020) , poverty (Ngono et al., 2023; Koudjom et al., 2022; Henao-Cespedes et al, 2022) , the informal sector (Nguimkeu & Okou, 2022) , population density (Nguimkeu & Tadadjeu, 2021; Koudjom et al., 2022) , and the population aged 65 and over (Nguimkeu & Tadadjeu, 2021) . Research has focused on the impact of COVID-19 on poverty (Ngono et al., 2023; Ehigie et al, 2021; Valensisi, 2020) , inequality (Ngono et al., 2023; Blundell et al., 2022; Dang & Nguyen, 2021) , digitalization (Amankwah-Amoah et al., 2021) , employment (Ngono et al., 2023; Su et al., 2022) , and resilience (Asongu et al., 2021; Diop et al., 2021) .
Although this literature is dense, to our knowledge no study has focused on measures to combat COVID-19 and resilience of economic growth by linking corruption. This is unfortunate, as studies show that COVID-19 has resulted in a loss of 0.3 to 2.2% of global GDP McKibbin and Fernando (2021). This slowdown in economic growth is also greater than that observed during the financial crisis of 2008-09. The worst outcome of the pandemic is the massive loss of output of 24.3% due to the high costs of health policies worldwide, (Acemoglu et al. 2021) . Sub-Saharan Africa, which had maintained an average growth rate of 4.5% between 2010 and 2019, recorded a negative economic growth rate of −1.4% in 2020, marking the first time in at least 25 years.
According to the AFDB report (2021)3 to slow the spread of the coronavirus, fewer than 42 African countries have imposed restrictions such as lockdowns, curfews, border closures, travel bans, and suspension of sports and recreational activities, all of which have impeded income-generating activities. Governments and international agencies have implemented various economic support programmes to mitigate the effects of the pandemic on vulnerable households and businesses (World Bank, 2020; UNDP Africa Regional Bureau, 2021) .
Stimulus packages have been implemented in order to safeguard impoverished households. However, numerous studies reveal that less than two out of every ten citizens and businesses received COVID-19 government aid in 2020 (African Union, 2020; Human Rights Watch, 2021a; International Labour Organization, 2021) . Although states have promised transparency in their use of COVID-19 funds, corruption, fraud, embezzlement, and lack of transparency regarding procurement processes, disbursements, and beneficiaries have plagued these aid initiatives (Human Rights Watch, 2021b; Oduor, 2021; United Nations Office on Drugs and Crime, 2020) . Afrobarometer surveys conducted across 16 African countries in 2020-2021 indicate that few citizens have reported receiving emergency financial aid offered by their government, with nearly 80% of citizens considering their government’s aid distribution unfair and suspecting that at least some of the funds allocated for COVID-19 response have been siphoned off through corruption.
Reducing corruption is target 16.54 of the 2030 SDGs (UNDP, 2015)5. Although often difficult to detect and measure, “corruption is an insidious evil with multiple and deleterious effects that affect every country in the world, and particularly developing countries, where resources that should be devoted to development are diverted6”. Indeed, according to data from the latest UN Secretary General’s report on the Sustainable Development Goals7, collected from 38 countries over the past 10 years, it appears that high-income countries have the lowest prevalence of corruption (3.7% on average), while low-income countries pay the most bribes to access public services (22.3%).
Thus, the aim of this project is to fill this gap in the literature by empirically analysing the effects of COVID-19 on the resilience of economic growth through the corruption channel in a sample of 11 Central African countries. Central Africa is the ideal field for such a study for several reasons: first, the area has been less contaminated but has the lowest growth rate - 4.4%; second, it is the region that has benefited most from IMF COVID-19 grants; and third, it is the region that hosts countries with considerable rankings on the list of the most corrupt countries (Human Rights Watch, 2021b) .
The study is based on the hypothesis that measures to combat COVID-19 contribute negatively to the resilience of economic growth through the mediated effect of corruption in Central Africa. Thus, the present study is built around the following question: what are the effects of measures to combat COVID-19 on the resilience of economic growth in Central Africa? From this question stems the objective of assessing the effects of measures to combat COVID-19 on the resilience of economic growth in Central Africa. Three subsidiary objectives emerge from this main objective, namely To assess the effect of COVID-19 control on the resilience of economic growth (direct effect). To assess the effect of the fight against COVID-19 on the recovery (return to equilibrium) of economic growth (direct effect). To assess the effects of measures to combat COVID-19 on the sensitivity (resistance) and recovery of economic growth with the mediated effect of corruption.
This study makes four contributions to the literature. First, unlike previous studies (Barro, 1991; Pongou et al., 2022) , which have focused on assessing the effect of COVID-19 on economic growth, this study highlights three points. First, the study models sensitivity and recovery indices for the period 2020-2021. Secondly, the study calculates the resilience and recovery and then assesses the effect of COVID-19 control measures on these indicators through the corruption channel. The second contribution is on the theoretical and empirical level. Indeed, to our knowledge, no study has assessed the effects of COVID-19 measures on the resilience of economic growth with corruption as a mediated effect. Therefore, this study contributes to the theoretical and empirical literature on the relationship between the three concepts.
Third, studies that deal with the resilience of economic growth in general analyse it in the long run (Briguglio, 2016; Talandier & Calixte, 2021) yet the work of Martin, et al. (2016) shows that it is better to analyse it in the short run. Fourthly, on the empirical level the study proposes the calculation of daily volume data starting from annual data. Also, the study proposes a three-legged methodology. The modelling of indicators, structural equations and the application of the N-SIRD model.
Generally speaking, the study is based initially on the modelling of sensitivity (resistance) and recovery (return to equilibrium) indicators. These indicators capture resilience in terms of resistance in 2020 and recovery in 2021. Therefore, it builds on the methodologies of Talandier & Calixte (2021) and Martin et al., (2016) . These indicators are modelled using a mathematical formalisation before being prepared for machine calculations. Similarly, the study is based on the Siefu (2022) structural equation model, and the N-SIRD mathematical model used by (Pongou et al., 2022) and will be estimated by the ordinary least squares (OLS) technique.
Wars, man-made disasters and catastrophes breed fraud and corruption at the same time as they enhance acts of selflessness and open-handedness. Openhandedness entails a lot, as individuals give not from the abundance of their pockets but from the pockets of institutions and as such bump up against institutions and practices designed for ordinary times. The concept of corruption is very complex and may at times go unnoticed. The COVID 19 pandemic was a deadly pandemic that claimed the lives of many people. Despite the devasting effects of this pandemic, it was a great opportunity for the corruption to thrive, where vaccines could be stolen, the theft of emergency funding and opportunities for nepotism, favoritism, and corrupted procurement systems. All of the above is frightening, given that during a health pandemic, qualities like self-sacrifice and love need to be at the fore front, but corruption is worse than any pandemic, because not only does it lag the economy behind, it leads to vicious cycle of poverty that will keep making a few rich and murdering the victims.
Due to the urgent requirement for medical supplies and protective equipment, the Healthcare Sector is particularly susceptible to vulnerability, resulting in the simplification of public procurement regulations. An April 2020 survey of anti-fraud professionals from 58 countries revealed a high prevalence of fraudulent practices in the procurement of personal protective equipment, black market merchandise, and defective equipment. Cases of embezzlement were reported by 19 percent of respondents from 58 percent of the surveyed countries, with uneven distribution. Three percent of respondents from 22 percent of countries reported instances of bribery. These reports demonstrate how corruption coexists with crises such as the COVID-19 pandemic. The unavailability of COVID-19 vaccines highlights the influence of external factors such as corruption. Sommer contends that corruption in the health sector diminishes the available funds for health, amplifies patient expenses, and hampers service improvement, reduces the quality of care and generally hinders reform and improvement (El-Gabalawy & Sommer, 2021) . Countries with health sectors that suffered from corruption prior to the crisis could be less effective during the COVID-19 pandemic.
As the COVID 19 pandemic was unfolding, there was an accompanying economic recession that led to fierce competition for essential resources. Governments therefore rapidly mobilized public funds (for both healthcare and for maintaining economic equilibrium) at an unprecedented scale, the government began creating opportunities for rent-seeking of many kinds, including outright corruption. This outright corruption was noticed as some governments scouted for funds, yet the proof of how these funds were fully put in use is not clear. Politicians, bureaucrats and medical professionals exercise substantial discretion in the allocation of resources. A lack of transparency and weak oversight and enforcement have exacerbated the problems of corruption and fraud, and public measures against these offenses have not kept pace with the developing crisis, making the international community to question the accountability of these pandemic regions.
Amongst the variables that can lag an economy behind is Pandemics. Evidence of this is how the COVID 19 pandemic wrecked several economies, distorting their development agendas keeping them away from achieving sustainable economic growth. The COVID 19 pandemic halted development and led to the stagnation of several economies. Several authors like Eichengreen et al., (2020) argue that there will possibly be long-term damages from a prolonged economic shutdown, should the pandemic persist. Bankrupt firms could disrupt the supply chains of surviving firms, and unemployed workers could lose their skills, competences and even their long-term relationships with firms (Apergis & Apergis, 2020) . Thus, having further negative effects on labor productivity and output growth.
Like disasters, epidemics can cause supply shortages, leading to destruction of production capacity and supply chain disruptions. Epidemics lead to disruptions in economic activity, spillover effects from production substitution, reconstruction needs, and all related paths to economic recovery (Kousky, 2016) . An epidemic is the sudden loss of a factor of production to which the economy adapts, whether to return to a pre-pandemic equilibrium or to transition to a new normal. A meta-analysis by Klomp & Valckx (2014) shows that the real negative impact on economic growth increases over time. Growth can be the result of technological progress, even if it is the destruction of human capital, as long as production supplies embody the latest methods of product innovation and adaptation to change (Caballero & Hammour, 1998) . Pandemic disasters can have severe impacts on human health (Kousky, 2016) and life satisfaction (Hudson et al., 2019) , all of which are fundamental elements necessary for economic growth. Bray et al. (2020) use the Spanish flu as an example to support concerns about the long-term effects of epidemics. There is no way to prevent new viruses from evolving and infecting humans (Martin & Pindyck, 2015) . Therefore, we are developing response strategies to understand that despite recurring pandemics such as COVID-19, economic growth should not be disrupted.
It is not a myth, that corruption is a threat to development and economic growth, corruption not only involves a few individuals, hoarding resources and accumulating them to their own advantage, while the suffering majority is left to wallow in poverty and to accept suffering as a way of life. There has been no agreement about the effect of corruption on economic growth. Some researchers suggest that corruption might be desirable (Leff, 1964; Acemoglu & Verdier, 1998) . This perspective however is very contrasting when we look at the devastating effects of corruption on the development indices of several countries today.
Mauro (2017) conducted an empirical analysis of corruption by examining the relationship between investment and corruption in 58 countries. Its corruption variable is defined as the extent to which business transactions involve corruption and questionable payments. The average ratios of total investment and private investment to GDP for the period 1970 to 1985 are from Barro (1991) , while the corruption indicator is a simple average across countries for the period 1980 to 1983 from Business International (1984). Mauro finds that corruption has a significant negative impact on the investment-to-GDP ratio.
Let p i b P 1 m , P I B P C m and p i b P 2 m as the growth rates at the peak P 1 m at the
trough C m and at peak P 2 m . Thus, I m r denotes the rate of decline of economic growth, i.e., its percentage decrease. It is calculated between P 1 m and C m (recession phase (r)), as follows:
I m r = p i b C m − p i b P 1 m p i b p 1 m
I m e The rate of decline of economic growth or the intensity of recovery of the gdp. It is calculated between points Cm and P2m (recession phase (e)) as follows:
I m e = p i b P 2 m − p i b C m p i b C m
From the resilience intensity variables, we can construct the sensitivity index (SI) which corresponds to the rate of decline of economic growth ( I m r ) of the Δ p i b m r relative to the national average decline.
I S p i b m = Δ p i b m r − Δ p i b n r Δ p i b n r
Also, we can construct the recovery index (RI) defined as the economic growth rate I m e of Δ p i b m e in relation to the average expansion of the country Δ p i b m e after the crisis the formulation:
R I p i b m = Δ p i b m e − Δ p i b n e Δ p i b n e
The following model allows us to evaluate the temporality of the shock in terms of both the decline and the recovery. For the duration of the decline ∆tC it is measured in years before the first peak P1m and the trough Cm with:
Δ t C = t C − t P 1
The rebound can be described as rapid when the duration of the decline calculated for each country is less than or equal to that of the reference cycle, i.e., 1 to 2 years. For the recovery time, ∆tR it is measured in years between the first peak P1m and the recovery point Rm8:
Δ t R = t R − t P 1
Recovery is said to be rapid when this duration is less than or equal to that of the reference cycle, which is 2 to 3 years, otherwise it is slow.
We deepen the study by conducting a mediation analysis. The study uses a mediation: corruption. The approach followed in this study is inspired by Ang (2013) and taken up by Ndoya et al. (2023) and illustrated in the
This approach involves the subsequent estimation of two regression equations which are:
Model 1: M e d i t = α 1 + β 1 s o u t i e n _ r e v i t + β 2 a l l e _ d e t t e i t + c ′ 1 X i t + μ i t
Model 2:
I n d i c a t e u r s r é s i l i e n c e = α 2 + β 1 s o u t i e n _ r e v i t + β 2 a l l e _ d e t t e i t + β 3 M e d i t + c ′ 1 X i t + v i t
Where Medit represents the mediation (corruption) variable. The composition effect is derived from the above two models as follows:
Indirect effect: b1*b3; direct effect: b2 and total effect: (b1*b3) + b2.
In a first step we estimate Model (1), which is the effect of the impact variables (income support and debt relief) on the mediator (corruption); b1 is the parameter describing this effect. The second step is to estimate model (2), where the resilience of growth is regressed through the sensitivity and recovery on debt support and the fiscal balance controlling for the mediator. The magnitude of this effect is provided by the coefficient on the income support and debt relief variables (b2). The indirect effect is obtained from the product of b1 and b3, where b3 measures the strength of correlation between income support, debt relief and the mediators in model (2). This term also reflects the size of the mediation, which essentially depends on the extent to which income support and debt relief affect the mediators (b1) and the extent to which the mediator influences the resilience of economic growth (b3).
Thus, the study adopts the N-SIRD model used by Pongou et al., (2022) . The equation of the model is the following:
c o v i d _ d e a t h i j s = α 0 λ s + α 1 E i g _ C e n t i j s + α 2 C o u n t r y _ s s a i j s + α 3 D _ P r o f i t i j s + β 1 λ s * E i g C e n t i j s + β 2 λ s * o u n t r y s s a i j s + β 3 λ s * D _ P r o f i t i j s + c ′ X i j s + θ j + ε i j s (1)
where c o v i d _ d e a t h i j s is a variable counting the total number of deaths due to Covid19 in nursing home i, county j and US state S; λ s is the tolerable incidence of infection in US state s; E i g _ C e n t i j s is the centralised eigenvector index for the nursing home; C o u n t r y _ s s a i j s is the average socio-economic status of county j ; D _ P r o f i t i j s is an indicator of whether nursing home i is for-profit (1 if for-profit and 0 otherwise); X i j s represents other exogenous characteristics of the nursing home, including the constant and θ j is the effet fixe of the county, α 0 ; α 1 ; α 2 ; α 3 and β 1 ; β 2 and β 3 are the parameters of interest. We take this linear equation from Pongou et al., (2022) by integrating our analysis variables. Thus, we will rewrite this equation in the simplest possible way as the index i will indicate the individual and j the period of application of the COVID-19 control measures. Also, we replace the dependent variable with our sensitivity indices (SI) in the first equation of 2020 and the recovery index (RI) in the second equation of 2021. In addition, the variables of interest are of two orders, the health measures and the economic measures; we keep the differences coefficients of interest. The following linear equations will be estimated.
S I _ p i b i j = α c + β 1 s o u t i e n _ r e v i j + β 2 a l l _ d e t t e c , t + β 3 i n d i c e _ c o n f i c , t + β 4 l a v _ m a i n c , t + β 5 f e r _ e c o c , t + β 6 f e r _ t r a v c , t + β 7 a n u _ e v e p u b c , t + β 8 r e s t r i _ r a s s e m c , t + β 9 c o m _ s e j d o m i c , t + β 10 r e s t r i _ c i r i n t e r n c , t + β 11 t r a n s _ c o m c , t + β 12 c o r r u p t i o n c , t + c ′ X i j s + θ j + ε i j s (2)
R I _ p i b i j = α c + β 1 s o u t i e n _ r e v i j + β 2 a l l _ d e t t e c , t + β 3 i n d i c e _ c o n f i c , t + β 4 l a v _ m a i n c , t + β 5 f e r _ e c o c , t + β 6 f e r _ t r a v c , t + β 7 a n u _ e v e p u b c , t + β 8 r e s t r i _ r a s s e m c , t + β 9 c o m _ s e j d o m i c , t + β 10 r e s t r i _ c i r i n t e r n c , t + β 11 t r a n s _ c o m c , t + β 12 c o r r u p t i o n c , t + c ′ X i j s + θ j + ε i j s (3)
These equations are estimated by the ordinary least squares (OLS) method (Pongou et al., 2022) . Other methods can be used to estimate these equations. For example, quantiles and generalized least squares. These methods are valid from the econometric tests
The study is conducted on a sample of 11 Central African countries over a period from 2020 to 2021. Thus, it uses daily data for the year 2020 and 2021 on anti-corruption measures. These daily data are derived from two sources, Oxford University9 which has been online since 1er January 2020. This database provides data on measures to combat COVID-19, specifically public health measures and economic measures. The Afrobarometer survey database on corruption will be online in 2020. In addition, the study uses World Bank data on economic growth rates over the periods 2020-2021. This data is used to model the sensitivity and recovery indices. In addition, after calculating these indices, they are made daily by the following formula:
V j = V c a l N j * 100
with Vj as the daily value to be obtained, Vcal as the global calculated value with the global data, Nj as the total number of days. With j ranging from 1 to 365 days. We multiply by 100 to get the estimable coefficients10. Secondary data on the variables used in the analysis were collected as follows:
Health policy measures taken as indicators. These indicators are 11 in number, namely, school closures, workplace closures, cancellation of public events, restriction on gatherings, public transport, home stay orders, restriction on internal circulation. These measures are categorised on a scale from 0 to 100 for some and 0 and 1 for others. We also have the hand washing index. The economic measures taken as indicators are also four namely, income support, debt relief. Also, income support measures. Economic growth is taken in terms of GDP11.
The analytical results reported in
The results presented in
Overall, the results suggest: (i) income support and debt relief affect the mediator and the effects are statistically significant at the 1% level (columns 1a, 2a). (ii) the mediator has a significant distinguishing effect on the resilience of economic growth (1b, and 2b). (iii) Income support and debt relief significantly affect the resilience of economic growth in the absence of the mediators (see
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | |
|---|---|---|---|---|---|---|---|---|
| VARIABLES | IS_gdp | IS_gdp | IS_gdp | IS_gdp | IS_gdp | IS_gdp | IS_gdp | IS_gdp |
| Income support | −0.0302** | −0.0528*** | −0.0314** | −0.0211* | ||||
| (0.0119) | (0.0126) | (0.0123) | (0.0125) | |||||
| Debt relief | −0.0580*** | −0.0955*** | −0.0922*** | −0.0873*** | ||||
| (0.00769) | (0.00762) | (0.00777) | (0.00794) | |||||
| Corruption | −0.383*** | −0.192*** | −0.149*** | |||||
| (0.0231) | (0.0262) | (0.0271) | ||||||
| School closure | −0.0987*** | −0.0831*** | −0.101*** | −0.0898*** | −0.0986*** | |||
| (0.0111) | (0.0103) | (0.0106) | (0.0110) | (0.0112) | ||||
| Closure of workplaces | −0.0312*** | −0.0427*** | −0.0626*** | −0.0366*** | −0.0549*** | |||
| (0.0102) | (0.00954) | (0.0107) | (0.00976) | (0.0109) | ||||
| Cancellation of events | −0.0737*** | 0.0665*** | −0.110*** | −0.0659*** | −0.0940*** | |||
| (0.0182) | (0.0170) | (0.0201) | (0.0170) | (0.0190) | ||||
| Quarantine | −0.0565*** | −0.0721*** | −0.0612*** | −0.0746*** | −0.0803*** | |||
| (0.00741) | (0.00740) | (0.00776) | (0.00731) | (0.00744) | ||||
| Containment index | −0.0448*** | −0.0530*** | −0.0315** | −0.0489*** | −0.0345** | |||
| (0.0139) | (0.0132) | (0.0152) | (0.0133) | (0.0146) | ||||
| Hand washing | −0.240*** | −0.218*** | −0.260*** | −0.216*** | −0.230*** | |||
| (0.0124) | (0.0123) | (0.0129) | (0.0123) | (0.0131) | ||||
| Restriction on assembly | −0.167*** | −0.183*** | −0.202*** | −0.183*** | −0.211*** | |||
| (0.0105) | (0.0104) | (0.0111) | (0.0105) | (0.0108) | ||||
| Public transport | −0.00983*** | −0.00615*** | −0.0112*** | −0.00595*** | −0.00686*** | |||
| (0.00106) | (0.00111) | (0.00106) | (0.00111) | (0.00111) | ||||
| Traffic restrictions | −0.0131*** | −0.0109*** | −0.0133*** | −0.0113*** | −0.0119*** | |||
| (0.000414) | (0.000400) | (0.000475) | (0.000407) | (0.000461) | ||||
| Constant | 0.169*** | 0.126*** | 0.148*** | 0.288*** | 0.324*** | 0.297*** | 0.329*** | 0.339*** |
| (0.00902) | (0.0103) | (0.00708) | (0.0250) | (0.0248) | (0.0260) | (0.0248) | (0.0255) | |
| Comments | 3,199 | 3,199 | 3,199 | 2,906 | 2,906 | 2,906 | 2,906 | 2,906 |
| R-squared | 0.002 | 0.017 | 0.079 | 0.428 | 0.448 | 0.434 | 0.450 | 0.455 |
Standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1. Source: Author.
| 1 (1a) | 2 (1b) | 3 (2a) | 4 (2a) | |||
|---|---|---|---|---|---|---|
| VARIABLES | Corruption | IS_gdp | Corruption | IS_gdp | ||
| Income support | −0.104*** | 0.0392*** | ||||
| (0.00727) | (0.0130) | |||||
| Debt relief | −0.0416*** | 0.0893*** | ||||
| (0.00441) | (0.00807) | |||||
| Corruption | −0.155*** | −0.178*** | ||||
| (0.0248) | (0.0262) | |||||
| School closure | −0.0946*** | −0.109*** | ||||
| (0.0100) | (0.0112) | |||||
| Closure of workplaces | −0.0598*** | −0.0535*** | ||||
| (0.0106) | (0.0112) | |||||
| Cancellation of events | 0.0956*** | 0.107*** | ||||
| (0.0174) | (0.0191) | |||||
| Quarantine | −0.212*** | −0.201*** | ||||
| (0.0108) | (0.0111) | |||||
| Containment index | 0.00703*** | 0.0107*** | ||||
| (0.00112) | (0.00110) | |||||
| Hand washing | −0.0117*** | −0.0137*** | ||||
| (0.000478) | (0.000431) | |||||
| Restriction on assembly | 0.0789*** | 0.0645*** | ||||
| (0.00741) | (0.00763) | |||||
| Public transport | −0.0366** | −0.0279* | ||||
| (0.0146) | (0.0153) | |||||
| Traffic restrictions | −0.232*** | −0.255*** | ||||
| (0.0127) | (0.0136) | |||||
| Constant | 0.141*** | 0.336*** | 0.144*** | 0.303*** | ||
| (0.00861) | (0.0246) | (0.00788) | (0.0260) | |||
| Comments | 2,906 | 2,906 | 2,906 | 2,906 | ||
| Mediation effect through corruption | ||||||
| Coeff. | Std. Error | P-value | Coeff. | Std. Error | P-value | |
| (A) Mediation test | ||||||
| Delta | −0.018 | 0.003 | 0.000 | −0.006 | 0.001 | 0.000 |
| Sobel | −0.018 | 0.003 | 0.000 | −0.006 | 0.001 | 0.000 |
| Monte Carlos | −0.019 | 0.003 | 0.000 | −0.007 | 0.001 | 0.000 |
| (A) Composition of the effect | ||||||
| Indirect effect (Sobel) | 0.4 | 0.054 | ||||
| Direct effect | 0.15 | 0.02 | ||||
| Total effect | 0.55 | 0.074 | ||||
| Of the total mediation effect | 32% | 7% | ||||
Source: Author.
Below the table, we report formal assessments of mediation effects using several statistical methods. Several mediation tests are considered to analyze whether the indirect effects of income support and debt relief on sensitivity through the impact of corruption are statistically different from zero. For example, if the mediating effect of corruption is taken into account, the Sobel test statistics are estimated to be −0.018 and −0.006 respectively. The P value is less than 5%, indicating that the null hypothesis of no mediation is rejected. Results were similar when using alternative mediation tests (Delta and Monte Carlos). It’s also worth noting that using bootstrap confidence intervals does not change the results. Indeed, the evidence presented shows that the mediating effect of corruption is significant, accounting for approximately 32% of the total effect of income support on the sensitivity of economic growth.
Overall, the measures to combat COVID-19 did not contribute to the recovery or the recovery of economic growth in 2021.
The results presented in
As in the first mediation analysis, the results suggest the same responses, i.e.: (i) income support and debt relief affect the mediator and the effects are statistically significant at the 1% level (columns 1a, 2a). (ii) the mediator has a significant distinctive effect on the recovery of economic growth (1b, 2b). (iii) Income support and debt relief significantly affect the recovery or return to equilibrium of economic growth in the absence of the mediator (see
Below
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | |
|---|---|---|---|---|---|---|---|---|
| VARIABLES | IR_gdp | IR_gdp | IR_gdp | IR_gdp | IR_gdp | IR_gdp | IR_gdp | IR_gdp |
| Income support | −0.227*** | −0.242*** | −0.0785*** | −0.0521*** | ||||
| (0.0230) | (0.0162) | (0.0165) | (0.0171) | |||||
| Debt relief | −0.190*** | −0.204*** | −0.185*** | −0.182*** | ||||
| (0.0121) | (0.00974) | (0.0102) | (0.0100) | |||||
| Corruption | −0.228*** | −0.330*** | −0.222*** | |||||
| (0.0298) | (0.0295) | (0.0298) | ||||||
| School closure | −0.143*** | −0.143*** | −0.180*** | −0.142*** | −0.166*** | |||
| (0.0113) | (0.0111) | (0.0132) | (0.0110) | (0.0124) | ||||
| Closure of workplaces | −0.134*** | −0.122*** | −0.131*** | −0.118*** | −0.104*** | |||
| (0.00950) | (0.00927) | (0.0102) | (0.00932) | (0.00978) | ||||
| Cancellation of events | −0.311*** | −0.319*** | −0.347*** | −0.319*** | −0.343*** | |||
| (0.0152) | (0.0147) | (0.0162) | (0.0147) | (0.0156) | ||||
| Quarantine | −0.214*** | −0.219*** | −0.187*** | −0.220*** | −0.205*** | |||
| (0.0101) | (0.00962) | (0.0104) | (0.00967) | (0.0102) | ||||
| Containment index | −0.0894*** | −0.126*** | −0.120*** | −0.122*** | −0.141*** | |||
| (0.0163) | (0.0166) | (0.0175) | (0.0166) | (0.0175) | ||||
| Hand washing | −0.0783*** | −0.0372*** | −0.0437*** | −0.0454*** | −0.0331** | |||
| (0.0125) | (0.0127) | (0.0134) | (0.0126) | (0.0131) | ||||
| Restriction on assembly | −0.183*** | −0.211*** | −0.245*** | −0.201*** | −0.226*** | |||
| (0.0114) | (0.0110) | (0.0115) | (0.0112) | (0.0112) | ||||
| Public transport | −0.00729*** | −0.00882*** | −0.00687*** | −0.00836*** | −0.00732*** | |||
| (0.000864) | (0.000843) | (0.000921) | (0.000850) | (0.000893) | ||||
| Traffic restrictions | −0.000638*** | −0.000654*** | −0.000679*** | −0.000662*** | −0.000712*** | |||
| (1.80e−05) | (1.73e−05) | (1.72e−05) | (1.76e−05) | (1.69e−05) | ||||
| Constant | 0.241*** | 0.297*** | 0.213*** | 0.676*** | 0.725*** | 0.602*** | 0.725*** | 0.686*** |
| (0.00985) | (0.0107) | (0.00898) | (0.0185) | (0.0183) | (0.0210) | (0.0184) | (0.0208) | |
| Comments | 4,015 | 4,015 | 4,015 | 4,015 | 4,015 | 4,015 | 4,015 | 4,015 |
| R-squared | 0.024 | 0.058 | 0.014 | 0.466 | 0.501 | 0.457 | 0.503 | 0.509 |
Standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1. Source: Author.
| 1 | 2 | 3 | 4 | |||
|---|---|---|---|---|---|---|
| VARIABLES | Corruption | IR_gdp | Corruption | IR_gdp | ||
| Income support | −0.142*** | −0.210*** | ||||
| (0.00728) | (0.0171) | |||||
| Debt relief | −0.105*** | −0.194*** | ||||
| (0.00525) | (0.00935) | |||||
| Corruption | 0.244*** | 0.237*** | ||||
| (0.0298) | (0.0293) | |||||
| School closure | −0.169*** | −0.168*** | ||||
| (0.0125) | (0.0123) | |||||
| Closure of workplaces | −0.118*** | −0.105*** | ||||
| (0.00961) | (0.00983) | |||||
| Cancellation of events | −0.338*** | −0.345*** | ||||
| (0.0158) | (0.0152) | |||||
| Quarantine | 0.211*** | 0.234*** | ||||
| (0.0114) | (0.0110) | |||||
| Containment index | −0.00616*** | −0.00754*** | ||||
| (0.000882) | (0.000871) | |||||
| Hand washing | −0.000694*** | −0.000711*** | ||||
| (1.67e−05) | (1.72e−05) | |||||
| Restriction on assembly | 0.198*** | 0.204*** | ||||
| (0.0106) | (0.0102) | |||||
| Public transport | 0.111*** | 0.145*** | ||||
| (0.0177) | (0.0173) | |||||
| Traffic restrictions | −0.0642*** | −0.0269** | ||||
| (0.0126) | (0.0123) | |||||
| Constant | 0.121*** | 0.634*** | 0.149*** | 0.683*** | ||
| (0.00606) | (0.0208) | (0.00726) | (0.0205) | |||
| Comments | 4,015 | 4,015 | 4,015 | 4,015 | ||
| Mediation effect through corruption | ||||||
| Coeff. | Std. Error | P-value | Coeff. | Std. Error | P-value | |
| (A) Mediation test | ||||||
| Delta | −0.035 | 0.005 | 0.000 | −0.025 | 0.003 | 0.000 |
| Sobel | −0.035 | 0.005 | 0.000 | −0.025 | 0.003 | 0.000 |
| Monte Carlos | −0.035 | 0.005 | 0.000 | −0.025 | 0.003 | 0.000 |
| (A) Composition of the effect | ||||||
| Indirect effect (Sobel) | 0.55 | 0.45 | ||||
| Direct effect | 0.20 | 0.11 | ||||
| Total effect | 0.70 | 0.56 | ||||
| Of the total mediation effect | 14% | 11% | ||||
Source: Author.
from zero. For example, when considering the mediating effect of corruption, the Sobel test statistics are estimated to be −0.035 and −0.025 respectively. The P value is less than 5%, indicating that the null hypothesis of no mediation is rejected. Results were similar when using alternative mediation tests (Delta and Monte Carlos). It’s also worth noting that using bootstrap confidence intervals does not change the results. Indeed, the evidence presented shows that the effect of mediating corruption is significant, accounting for approximately 14% of the total impact of income support and 11% of the total impact of debt relief on the recovery or return to equilibrium of economic growth.
While there is a large and growing body of literature that has analysed the effects of COVID-19 on economic growth, the relationship between government responses and the resilience of economic growth is not known. The objective of this paper is to fill this major gap. Using a sample of 11 Central African countries for a period 2020-2021 (daily case) and applying the ordinary least squares method and the mediation model, we test three hypotheses. First, COVID-19 control measures reduce the sensitivity of economic growth in 2020; second, COVID-19 control measures reduce the resilience of economic growth in 2021; and third, COVID-19 control measures reduce the sensitivity and resilience of economic growth in the presence of the mediator (corruption).
Overall, our results show that COVID-19 control measures (income support, debt relief, school closures, workplace closures, cancellation of public events, quarantine and containment) reduce the ability of economic growth to withstand the health shock of COVID-19 and to return to equilibrium after the shock. This result corroborates that of Hu & Liu (2022) who showed that COVID-19 reduces the ability of large metropolises to withstand shocks during COVID-19. Finally, we conduct a transmission channel analysis and without being exhaustive, we find that corruption is a channel through which COVID-19 control measures affect the resilience of economic growth.
From a political point of view, as states and organisations seek sustainable solutions to economic, health and other crises, despite the measures taken during the COVID-19 health shock, economic growth is struggling to resist or to recover instantly. Therefore, we suggest a strengthening of political, economic and health institutions to fight the next shocks. Also, the re-education of the authorities in policy decisions in the presence of a shock. Similarly, the authorities need to learn from previous crises and, above all, go back and read the solutions adopted by the major economies during the shocks before 1950.The study obviously leaves room for future research, particularly with regard to the consideration of the corruption channel.
Not applicable.
Not applicable.
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Emmanuel, O. N. B., & Merlin, N. A. (2024). COVID-19 Control Measure and Economic Growth Resilience in Central Africa: Does Corruption Matter? Voice of the Publisher, 10, 1-20. https://doi.org/10.4236/vp.2024.101001
AFBD: African Developpment Bank
GDP: Growth Domestic Product
IMF: International Monetary Fund
N-SIRD: Population-Susceptible-Infected-Recovered-Deceased
OLS: Ordinary Least Squares
SEM: Structure Equation Model
UN: United Nations
UNDP: United Developpment Programme