Capital Flight, Hard Mineral Resources Exports and Political Governance Crisis in the Central African Republic: How Far We have Learned?

The Autoregressive Distributed Lag (ARDL) approach to cointegration is applied to examine the role of political governance crisis in the relationship between hard mineral resources exports particularly diamond, and gold and Capital ight in the Central African Republic over the period of 1978-2010. The results of short-run show that ocial development assistance, gross national expenditures have a negative and signicant impact on capital ight while the interaction of political governance crisis and hard mineral resources, GDP per capita have a positive and signicant effect with capital ight. Moreover, governance crisis, exports of hard mineral resources has a negative and insignicant impact on capital ight. The results of the long run show a negative and signicant effect on governance crisis, ocial development assistance and gross of national expenditures. The interaction term between hard mineral resources and GDP growth have a positive and signicant effect on capital ight. However, export of hard mineral resources exhibits a positive but insignicant effect on capital ight. Without strong action to stop and promote political governance stability and the implementation of appropriate reforms at institutional level in CAR’s hard minerals resources sector, success in the ght against capital ight is unlikely.


Introduction
Capital ight has become an increasing source of concern for political actors in developed countries, such as the Central African Republic (CAR), where there are insu cient resources to nance growth. According to Ajayi (1997), capital ight has been attributed to slow economic growth and persistent balance of payment de cits in most developing countries.Indeed, high capital ight rates pose signi cant challenges to the mobilization of domestic resources to sustain investment and development in Africa Ndikumana, 2009 and2010).With respect to Boyce and Ndikumana (2012), CAR is ranked 26th among the sub-Saharan African countries with the highest capital ight rates in the 1970-2010 period. The total real capital and total real capital ight per capita were estimated at $ 2.7 and 619.2 billion 2010 $ versus Cameroon estimated at $20 and $1020 billion2010 $. While the volume is low compared to other African countries such as Cameroon, careful attention should be given to the fact that it poses a comparatively heavier burden (Nkurunziza, 2015) and that the country still faces major problems in terms of growth, including many off-track sustainable development goals (SDGs).The country's GDP per capita was USD 489.87 from 1960 to 2017, which is equivalent to 3% of the world's average GDP per capita and is considerably lower than that of other fragile countries in sub-Saharan Africa. In fact, its population continues to suffer from severe poverty (76% of the central African citizens live of less than US$ 1.0 a day).In addition, the political governance crisis that the country has experienced since its independence has exacerbated the situation and created tensions over hard mineral resources, leading to poor governance of natural resources. As long as bad governance has been reported as an signi cant determinant of capital ight through trade mis-invoicing, corruption, tax evasion, rent-seeking (Le Billon, 2011;Ljungberg and Friedl, 2014), a strong link between natural resources and capital ight may be identi ed through illicit offshore transfer and subsequent deposit of illicit funds in tax havens. (Asongu and Odhiambo, 2019).
While the literature on capital ight has recently centred on the institutional climate (Gankou et al. 2016), as well as the effect of governance on capital ight (Asongu and Nwachukwu, 2017, Yapatake and Ngaba, 2019), a broad body of literature that explored the relation between abundance of resources and con ict (Le Billon, 2011;Ross et al., 2013) has provided two contradictory ndings. Some researchers described natural resources as a blessing (Arezki and Van der Ploeg, 2007) and others as a curse (Egorov et al., 2009;Polterovich et al., 2007;Kolstad, 2009). In addition, the capital ight literature has also highlighted the role of political risk and political governance crisis as factors in driving capital ight (Yapatake and Njong, 2020; Lensink et al., 2000;Le and Zak, 2006). Furthermore, there is evidence in the literature that African countries with rich natural resources seem to be prominently at the top of the list of countries with high capital ight, and the correlation seems stronger for oil than other resources such as hard mineral resources (Ndikumana et al. 2016). Nevertheless, the increasing number of work on governance (Kangoye;Musila and Sigué, 2010) and capital ight (Mpenya et al. 2015;Ndiaye and Siri, 2016) lack of literature illustrates the role of the political governance crisis in the relationship between capital ight and hard mineral resources.
The Central African Republic is regarded as a resource-dependent nation and an insecure democratic stability region marked by civil wars, protests, coups d'états, inter-confessional con icts, the continuing involvement of foreign forces that tend to exacerbate the capital ight. Studies also report that maladministrated resources create grievances and increase probability, duration and severity of con icts (Valenzuela, 2020;Vogel andHavenith, 2017, Aspinall, 2007). With an insecure political climate and a global governance crisis, countries rich in natural resources were faced with badly de ned land rights, fragmented economies, poorly structured political institutions, instability and insecurity. These conditions offer the best incentives for corporations, policymakers and senior o cials to seek rent that diverts money from more e cient economic projects (Auty, 2004;Le and Zak, 2006).
In interpretation of the political governance crisis as one of the dimensions of poor governance, this study empirically examines the role of the political governance crisis in the relationship between capital ight and hard mineral resources in the CAR due to the fact that CAR is not spared from the problem of capital ight and the ght against capital ight, which could therefore be an important catalyst of the country's development. As far as I know, no substantial research has been carried out on capital ight, hard mineral resources and the CAR governance crisis in particular. One wonders, though, if capital ight is not a major issue for the Central

African Republic
The uniqueness of this study is therefore to provide a detailed evaluation of political governance through an index construct that speci cally takes into account changes in constitution and government, tensions, violence and politically motivated, as well as political-military crises, as a factor leading to the variance of capital ight in hard mineral resources.

Hard mineral resources and Political governance crisis in the Central African Republic
The CAR is endowed with rich natural resources such as diamond, gold, uranium, iron, and tin, copper and nickel (see Fig. 1). It is from 1920 that mining prospects were undertaken, in particular by the Equatorial Company of the Mines in the Center and the East of the country.A law of January 17, 1961, gave the right to Central Africans to open their small mines for gold and diamonds. Mainly artisanal workers carry out the gained its independence, there were efforts to start large-scale gold mining but persistent civil strife slowed down the growth of the country in the mining sector. The causes of the political governance crisis in the CAR are diverse and varied. Military coups, army mutinies, and extended labor disputes, combined with poor macroeconomic management and weak governance, have led to the deterioration of the political environment as well as economic and social performance. In addition to that, road blockers or Zaraguinas, which are located mainly in the West, the North-west, and the Northeast and in the Center of the country, have seriously contributed to the degradation of the internal security. Another possible cause of the political governance crisis is linked to the extreme poverty that prevails throughout CAR, especially the high socio-economic disparities between the north and other regions. As well as the geopolitical position with other countries in particular Chad, Sudan, the Democratic Republic of Congo has contributed partly to the destabilization of the country.
The weakness of the national defense is also a factor of political governance crisis and the limited resources of Central African Army adding to those natural resources rent-seeking. Besides that, the country is characterized frequents changes in the constitution and government changes which are always conditioned by abrupt changes in executive power .The country was also shaken by military crisis, punctuated by coups d'état, mutinies, civil wars, interethnic and confessional con icts, rebellions, regional instability as well as French support for dictatorial regimes. About 60 percent of the territory remains controlled by armed groups and the country still experienced an upsurge in clashes between armed groups, competing for territory, and access to resources. Each episode of violence brings new civilian casualties and additional displacement.Tensions, violence, and protestations politically motivated have shared the political history of the country and have created deeper divisions within the population, leading to an increase in the likelihood of continued violence based on identity.
The capital ight literature has explored some causes of capital ight including GDP growth, political instability, corruption, institutional quality (Bredino et al., 2018;Yapatake and Ngaba, 2019;Gankou et al., 2016). Previous research on the connection between natural resource and capital ight is not limited to a study by (Demachi, 2013;Arezki et al., 2014;Kwaramba et al., 2016). In addition to that, the study conducted by Ndikumana and Sarr (2019) found high natural resource rents are associated with high capital ight and the quality of institutions does not mitigate this link. This means also that a signi cant amount of rentseeking takes place in resource economies, possibly fuelling capital ight.
In light of the objective, the study uses an index developed by us in order to highlight country peculiarities with regard to capital ight and hard mineral resources in the context of a political governance crisis. Several proposed effects of hard mineral resources and explanatory variables for capital ight coincide. The implication is that there should exist an observable correlation between the interaction of hard mineral resources and political governance crisis in relation to the capital ight considering the political governance crisis as the main channel by which this correlation occurs. The construction of the index is motivated by the fact that the Central African Republic has experienced an unstable political governance crisis which however has not been studied in relation to hard mineral resources and capital ight.

Capital Flight and Natural Resources: Overview of the Transmission Mechanisms
Numerous studies have looked at the capital ight determinants. (Ndikumana, 2012;Lonescu, 2013) have considered low investment in infrastructure, poor rule of law, budgets de cits and foreign borrowing as drivers of capital ight. Several studies have also outlined corruption, administrative e ciency, tax revenue collection, export misinvoicing, export mispricing (Ndikumana and Boyce, 2003;DeBoyrie, 2011;Mpenya et al., 2016;Ljungberg and Friedl, 2014). Speci cally, by using exports as total share of total exports as proxy of natural resources, Ondo and Taylor (2012) nd natural resources are directly related to the ight of natural capital.
There was also an emphasis on the effect of political instability and poor governance in the induction of capital ight (Ramiandrisoa and Rakotomanana, 2016;Ndiaye and Siri;2016). In order to discuss the relationship between capital ight, hard mineral exports and the CAR political governance crisis, we operationalized the concept of capital ight and hard mineral resources exports and political governance crisis by constructing a three-pronged ow chart.
Throughout the rst phase of capital and natural resources, Boyce and Ndikumana (2012) clearly show that the sudden rise of capital ight in sub-Saharan African countries are correlated with a resource extraction growth during the pre-crisis period particularly in the main oil-producing countries. In particular, given that the CAR is well endowed with natural resources, a better understanding of this aspect is based on the effect of capital ight on the country's hard mineral revenues.
In the context of the CAR, where hard mineral resources are mined almost exclusively by craft miners and diamond workers, there is a high risk of corruption and a heightened impact of capital ight, notably through illicit exploitation and export smuggling, given the large number of informal operators. In addition, the capital ight in hard mineral sector is basically illustrated by the high value / weight ratio and the relative ease with which they can be produced, resulting in contraband, trade misinvoicing, corruption, illicit mining, tax base reduction depriving the country of signi cant amounts of revenues.
As regards the second aspect, the link between capital ight and the political governance crisis can be examined from the point of view of political risk as a determinant of capital ight. Boyce and Ndikumana (2012) have shown that bad governance due to abuse of political power is likely to be correlated to capital ight. Furthermore (Gibson and Tsakalotos, 1993;Ndikumana et.al, 2015;Davies, 2008 andFatehi, 1994) showed that political risk and a weakening market climate were signi cant determinants of capital ight through damages and asset loss. Our CAR focus has enabled us to notice that the country is marked by an on-going political turmoil due to civil wars, coup d'états, minuties, changes in government, constitutional reforms that weaken the foundations of stability in democratic governance (Yapatake Kossele and Shan, 2018).
Provided the above-mentioned investors' reaction to the country's political events by moving resources to environments with lower investment risks (Hermes and Lensink, 2001;Le and Zak, 2006). Several scholars have also suggested in the third aspect on the natural resources and political stability issue that countries with an surplus of exportable point-source natural resources ( e.g. crude, natural gas, copper) are more likely to be ruled by oppressive regimes (Ross, 2001;Wantchekon ,2002). However, it could quickly suffer from economic stagnation (Bova et al., 2018) and is frequently driven by violence and political uncertainty (Ong'ayo, 2008;Williams and Le Billon, 2017). Supporting a foreign law that encourages rulers and greedy politicians to assume legally legitimate property transactions leads to irresponsible administration, natural resource mismanagement, tax haven, and poverty in many African countries such as the CAR.
In the context of the study's purpose, the relations between capital ight, hard mineral resources and political governance crisis can be explained from the viewpoint of how the political governance crisis affects capital ight movements in the diamond sector. Figure 3 shows the connection between capital ight and hard mineral resources; the connection between diamond exports and the crisis of political governance; and nally the interaction between hard mineral resources exports and the crisis of political governance in relation to capital ight. Considering the role of the political governance crisis in the relationship between capital ight and hard mineral wealth, the framework may be formed by growing instability, which would drive investors to take their investments somewhere else. The political governance crisis may lead to an over-estimate or under-estimate of the invoice, as well as the price of commercial transactions involving natural resources, the distortion of the amount of transactions making it di cult for the invoice sent to customs to represent the real cost of the transactions.This statement should be especially veri able in the CAR as a consequence of the political governance turmoil that the country has been going through continuously since its independence in 1960.

Methodology
The empirical methodology of this article applies the ARDL bounds test as proposed by Pesaran and and Smith (2001) to investigate the long-and short-run impacts of the selected variables on capital ight in the CAR over the period of 1978-2010. Therefore, the empirical methodology follows the next steps: First, checking for the unit root using the Augmented Dickey-Fuller (ADF) and Philips Perron tests for the selected variables. Second, based on the empirical analysis of stationarity, we will apply the ARDL bounds test for cointegration in the presence of structural breaks to capture the effect of hard mineral resources and political governance crisis and other variables on capital ight.
One of the reasons for preferring the ARDL is its applicability irrespective of whether the underlying regressors are purely or mutually cointegrated. Besides, endogeneity problems are addressed in this technique. According to (Pesaran and Shin, 1999), modeling the ARDL with the appropriate lags will correct for both serial correlation and endogeneity problems. Jalil et al (2013) argue that endogeneity is less of a problem if the estimated ARDL model is free of serial correlation. In this approach, all the variables are assumed to be endogenous and the long run and short-run parameters of the model are estimated simultaneously (Khan et al, 2005). Moreover, using the ARDL approach is more robust and performs better for small sample sizes than other cointegration techniques. The data of capital ight is obtained from Boyce and Ndikumana (2012).  January 18, 1978, beginning of the crisis of con dence in the army and suspicion of intellectuals and even students and students, having claimed the payment of salary arrears of their parents. Central African pupils and students have been beaten.
January 1979: in January, the Imperial Guard killed young people who protest against the high price of the uniforms imposed by Bokassa. Farmers also have been brutalized for protesting against a rise in food. One of the notorious violence during this year was about captain Alexander Banza who was mutilated before being dragged into the street and then executed. The following model is used to under the rationale behind the relationship between capital ight, hard mineral resources and political governance crisis.
To examine the role of political governance crisis in the relationship between capital ight and hard mineral resources, accent has been put on seven independents variables. Taking into the previous literature on the transmission mechanism of capital which occurs with the country, the choice of our variables is carefully done based on the previous literature and the adequacy of data to be sought.
Based on critical examination and review on capital ight, countries with an unstable political and weak institution are supported by macroeconomic and nancial instability. Ramiandrisoa and Rakotomanana (2016) by undertaking a detailed review from Madagascar's political and economic history with the aim to explain the uctuations of capital ight over 1970-2010 found that political cycles and crises are key determinants of capital ight. In similarity to this work, we expect a positive link between the interaction of hard mineral resources exports, capital ight and governance crisis in CAR as well the positive relationship between hard mineral resources exports, governance crisis, and capital ight.
In regards to o cial development, its links with capital ight are not conclusive. The study conducted by Gankou et al., (2016)  In regards to GDP per capita and GNE, we expected a positive link with capital due to their negative outcomes highlighted in Sect. 1. Moreover, the governance crisis and hard mineral resources have interacted in the model in order to test the hypothesis that the relationship between the amounts of hard mineral resources exports is more signi cant than when they are separated. Author's construction

Results And Discussions
The Table 2 presents the summary statistics for all variables used in this study over the period of 1978-2010 in CAR. For instance, the mean of KF for the sample is 62.27931 per year, with a standard deviation of 89.14983, which varies signi cantly. Hard mineral exports have a minimum of 9506.286 and a maximum of 9.26E + 08. The average is 2.96E + 08 and the standard deviation is 2.85E + 08, which is large, meaning it exhibit more dispersion from the mean. The interaction between governance crisis and hard mineral resources has the lowest value of 0.000000 and the highest value of 9.26E + 08. It has an average of 2.08E + 08 and the standard deviation is 2.73E + 08 which exceed the mean, hence it exhibit from the mean. Moreover, the Kurtosis values and Jarque-Bera test statistics indicate our variables are normally distributed and positively skewed.  The optimal lag order for ADF test is determined by AIC, while the bandwidths for PP ***Signi cance at 1% level; **signi cance at 5% level; *signi cance at 10% level.
Before conducting the regression model, it is necessary to check whether our variables are stationary. Moreover, The ARDL method is only applicable for times series which are combination of I (0) and I (1) variables. Thus, the test of stationarity is an imperative step in our empirical analysis. ADF and PP unit roots are applied to con rm whether our variables have unit root problem. As can be observed in the Table 3, the results of ADF at the rst level are mixed. ADF tests statistics accepted the unit-root null hypothesis against the stationarity only for ODA. However, in the rst differences, all the times series rejected the unit root null hypothesis at 1, 5, and 10% signi cance levels, which means there is homoscedasticity. The non-presence of stationarity allows us to examine the long run relationship between capital ight and Hard mineral resources by using ARDL approach to cointegration. Before apply the bounds test cointegration and select the suitable ARDL model, Akaike information criterion (AIC) is applied to choose the adequate optimal lag. The criterion for variables lag order section is presented in the Table 4. By minimizing the AIC value, it can be observed that the maximum of lags is set to 1.  The interaction between hard mineral resources and governance crisis, GDP per capita have a positive and a signi cance effect on capital ight. This positive link could be explained by the duration of the governance crisis. Once, for example, such as features are associated with poor regulation in the hard market, political uncertainty and crime, investors are likely to move their resources to less risky environment. The nding is in line with the work of Ndiaye and Siri (2016) who found a positive link between capital ight and natural resources as well as the study of Mpenya et al., 2016. A hard mineral resources export exhibits a positive but insigni cant effect on capital ight.
In the short-run analysis, the results show that o cial development assistance, gross national expenditures have a negative and a signi cant impact at 5% and 1% signi cance level on capital ight. Speci cally, a onedollar increase in o cial development assistance and gross national expenditures decrease the capital ight by -3.05, -7.29 dollar respectively. The interaction of hard mineral resources and governance crisis and GDP per capita has a positive and a signi cant effect at 5% levels of signi cance. The relationship is such that a one dollar increases in interaction between governance crisis and hard mineral resources and GDP per capita leads to 2.07, 2.04-dollar increase in capital ight.
Governance crisis and hard mineral resources have a negative but insigni cant impact on capital ight. The error correction model which shows the speed adjustment parameter and determines the pace of adaptation towards the long-term equilibrium is negative and signi cant but does not comprise between 0 and − 1 for our model. This suggests that the discrepancies between shocks and the trend are reduced in less than one year. In such cases, equilibrium is achieved by uctuations of capital ight, with the amplitude getting smaller until the nal extinction of the shock (Narayan and Smith, 2006).
To ensure the goodness of t of model, the diagnostic and stability tests are performed. For the diagnostic test, Jarque-Bera test for normality and the Breush-Godrey test for serial correlation, Breusch-Pagan-Godfrey for heteroskedasticity have been examined. For the stability tests, Cumulative (CUSUM) and cumulative sum of squares (CUSUMsq) have been conducted for goodness of t. The results in Table 6 show that our model has no serial correlation, normally distributed and free from heteroskedasticity. Moreover, the result of the stability test show that the plots of CUSUM and CUSUMsq fall with the critical bounds of 5% level of signi cance. This implies that, our model is well speci ed and stable over the period of 1978-2010

Conclusion
This study has examined the relationship between capital ight, hard mineral resources exports and capital ight in Central African Republic for the 1978-2010, focusing governance crisis index constructed on detailed review of the country's governments and constitution changes, Tensions, violence and protest politically motivated as well as politico-military crises.
The ARDL approach to cointegration is applied to identify long run and short run dynamics between selected variables. The results of short and long run show that the association of hard mineral resources and governance crisis have a positive and a signi cant relationship at 5% levels of signi cance while the rest of variables under our investigated have a mixed relationship in short and long run with capital ight in CAR.
Considering this signi cant level of capital ight, hard mineral resources and governance crisis, this paper has several policy implications. First, the country needs a rigorous reform in security sector and democratic governance in order to have an effective centralized, hierarchical, coercive government. Second, support peace and reconciliation by strengthening the performance of the security and justice sectors, and improving democratic governance and electoral processes. Third, there is evidence of hard mineral resources fueled capital ight in CAR, therefore there is a need to put in place a consensual and incentive management rules to promote transparency in the management of hard mineral resources in CAR. Moreover, there is a need to strengthen the human and material capacities of public bodies in charge of the management of the sector (General directorate of mines, Kimberley process, the Special Anti-Fraud Unit (USAF) to avoid corruption and trade misinvoicing considered as determinants of capital ight.