The Dynamic Relationship between Oil Rent Corruption and Political Stability: An Empirical Study on Determined Sample of African Countries Jilani Ikbal, Doctor in Economic Sciences, Research unit URECA FSEG Sfax, Tunisia. E-mail: kan1980@live.fr Abstract In this paper, we will examine the direct effects of the oil rent on the corruption and the political stability by exploiting sample of 17 African countries from 2000 to 2014 using a dynamic vector error correction model. We will find that the increase of the oil revenues generates an increase in the corruption index and results in a significant deterioration of political rights unparalleled. Indeed, this can be explained by the fact that the political elite abusing its power to divert the oil flow considerably his own. The leaders including talking bad decision for lack of control and a purely autocratic system created by themselves. Key Words: oil rent, corruption, political stability 1
1. Introduction During last decade. The natural resources have been elucidated through the political economy, the oil revenues have direct effects on economic policies and governance mode. It leads to rent-seeking behavior by political and military elites and opens the door to corruption. A significant portion of rent is regularly diverted from official and natural resources, and the exploitation of natural resources society. The oil resources affect the political democracy. Governments tend to use their oil revenues to reduce the social pressure and prevent the protest groups which could seek political rights, the oil rent can even cause an extreme case of institution collapse. Numerous oil exporters are indeed marks by social conflicts and structural policies that generate situations of violence. The majority of net exporters of oil depend on heavily dependent on oil revenues, indeed a greater part of the state revenues comes from the oil sector through the license fee, taxes on oil companies and signing bonus: from the FMI 80% of tax revenue of the governments Angola and Nigeria come from oil industry such income is paid directly by oil to government. The corruption usually finds fertile ground in extractives industries where the extraction of natural resources requires a high investment capital and an important technological knowledge which puts developing countries rich in natural resources dependent on the foreign companies to operate this kind of investment. These factors represent opportunities for the use of bribes between multinationals and the local leaders. As we said previously, the extractive industries offer numerous factors that promote corruption. Briefly, it is difficult to detect corruption in this type of industries. This can be explained by the natural and the way of payment which is not declared, nontransparent and are made through a complicated chain of intermediate without government control or state power. Besides oil revenues are largely paid in foreign currency. It can easily supply offshore secret accounts. Finally, crude prices fluctuations facilitate false declarations of oil revenues and mask the actual value of transactions. 2. Literature Paper A big number of authors whether in political or in economic sciences agreed on the fact that oil rents are related to corruption and the stability of the state. More generally, countries dependent on oil are often characterised by the corruption and exceptionally poor governance. The natural resources are a curse rather than a blessing may be seen paradoxical at first and leads to a wide literature. One of the most remarkable findings is that the natural resources supply the corruption, which in return, decreases the economic performance. The studies directed at this issue of Ades and Ditella (1999) confirm what was just said. With the help of a theoretical model and the two authors show that the revenues of natural resources lead to bureaucratic corruption, the proof is that the corruption index increases proportionally to 2
mining or oil exports volumes of metals. Leite and Weidman (1999) formed that the national resources affect the corruption on a nonlinear manner likely and that it is valid only when the natural resources are of oil or mining origin. Ross (1999) examined the political aspects the reason for which the countries rich in natural resources tend to mismanage their economies by showing that they lead politicians to abuse their political power for personal gain Treisman (2000) in his study shows the robustness of this proportionality. In another study done by Leite and Weidman (2000), we see clearly that any increase in natural resources export (in PNB percentage) tends to increase the corruption and any corruption increase made decreasing growth. Collier and Hoeffler (2005) deepen their analysis of a political system of oil economic their results suggest that when the natural resources provide significant revenue to governments, the performance of policies of autocrats are better than those of democrats. O Higgnis (2006) stresses that the corruption in the extractive industries in often considered as systematic according to Riely (1998) classification. The rent-seeking behaviors are often of the highest level, particularly when licensing or business operating license. Aslaksen (2006) worked on a sample of countries which is subdivided into three groups according to the degree of measuring its index of polity 2 (low, average and high) that measure the corruption in 1982. She concluded that natural resources, especially the mining and oil ones, increase the corruption. Collier on a sample of determined countries (developing countries) and by using a panel regression model over a period from 1970 to 2001 found that coupling of oil rents and open democratic systems ends at reducing growth. Likewise, Bhattacharyya and Holder (2010), Tsin (2010), Haber and Menaldo (2011) found similar results, their studies are usually based on measurements of resource rent taking into account the changes in international commodity prices and changes in the quantity of the extracted commodity and extraction costs in the time. For most countries, variations in commodity prices are the origin of variation in resource rents. It seems logical to conclude that all changes in political institutions or the degree of corruption reflect changes in the quantities of extracted resources. Finally, Haber and Menaldo (2011) find that oil contributes to the submission and obedience of the society. As well, the oil money, generated by tiny part of the population, can easily be diverted to the countries that have delicate institutions. The oil resources don't only represent a fertile ground for corruption but also it establishes, in particular, an obstruction to the democracy and the political liberty. Generally, the oil-dependent countries are often characterized by corruption and exceptionally a poor governance. 3
3. Modeling In this model we examine more closely the relationship between natural resources and corruption, in particular, the effects of oil rents on corruption and the stability of states rich in natural resources, we will show by means of a dynamic error correction model how the quality of democratic institutions affect this relationship by exploiting a sample extending over 17 oil exporters during a period from 2000 until 2014 we find that lower oil rents lead to a significant decrease of corruption as well as a deterioration of political rights. We support that these results can be explained by the political elite who have the incentive to expand the civil liberty and reduce the political rights in presence of richness coming from oil and escape redistribution and conflicts. 3.1 Variable Data Our oil rent indicator is the unit value of oil exports. Specifically, we rely on the value of export of oil per unit collected through surveys done by the IMF. The unit value of oil export is constructed using international oil prices affected by a specific factor of reduction of the countries that capture the quality of any given country. The capacity of production and of the oil quality are determined by geological exogenous factors, in turn, determine the chemical proprieties of oil (as the viscosity, sulphur, and the acid level), that determine finally the price by which oil can be sold in the international competitive market. The unit value of oil exportations is available for 17 countries which produce oil the period from 2000 to 2014. Therefore, the specific geological factors of each country influence the specific oil rents of each of these countries and then affect the unit price of the sale. The interested countries of study are Algeria, Angola, Equatorial Guinea, Guinea, Nigeria, Libya, Republic of Congo, Democratic Republic of Congo, South Africa, Thad, Egypt, Tunisia, Cameroun, Ivory Coast, Mauritania, Ghana, Maroc, over the period from 2000 to 2014. Oil production is measured by the production of crude oil, LNG and other liquids (as the ethanol and the biodiesel) in thousands of barrels per day drawn from the database of the World Bank. (2005). The corruption is measured by the index of polity 2 according to Marshell and Jaggers The polity 2 score is based on the constraints imposed on the chief Executive, the competitiveness of political participation, of the opening and of the competitiveness of the election score. The corruption index ranges between 10 and -10, the highest values are the strangest the democratic institutions are and vice versa. In order to avoid the negative values of this index we applied the following formula: (cc + 10) CC = 2 4
The last variable is the PIB: specifically the price buyers PIB which is the sum of gross value added by all resident producers in the economy. Overall taxes on products and minus any subsidies not included in the value of production for depreciation of fabricated assets or impairment or degradation of nature resources. The data are in current US dollars. Dollar amounts are converted to PIB from local currencies using the official exchange rate in one year. For some countries where the official exchange rate does not represent the actually applied to foreign currency rate transactions. An alternative conversion factor is used. Data on PIB are taken on the basis of IMF site data. 3.2 Methodology and Description of Data Table 1: Description Statistics PIB P2 PP RP Mean 65.43787 4.272124 513.5265 19.97920 Median 26.86300 3.500000 117.0000 7.600000 Maximum 521.8120 9.500000 2627.000 203.0000 Minimum 1.243000 1.000000 0.000000 0.000000 Std. Dev. 92.09183 2.405562 722.6606 25.31234 Skewness 2.528697 0.794890 1.418795 2.316173 Kurtosis 9.746361 2.384490 3.658742 13.74426 Jarque-Bera 669.4367 27.36723 79.90848 1289.120 Probability 0.000000 0.000001 0.000000 0.000000 Sum 14788.96 965.5000 116057.0 4515.300 Sum Sq. Dev. 1908204. 1302.014 1.18E+08 144160.7 Observations 226 226 226 226 The above table summarizes the descriptive statistics associated with the four variables. PP, P2, PIB and PR that are downloaded from the MFI indicators from 2000 to 2014. The empirical survey is based on 226 annual observations from the sample countries. All variables have a positive value of asymmetry (Skewness), which indicates that the distribution is biased to the right, with several observations on the left. Jarque-Bera statistics show that all variables used in the analysis have a normal distribution. 5
3.2.1 The Results of the Unit Root Tests To check the stationarity properties of the variables we use the following tests: Levin tests and al (2002) (LLC), Im and al (2003) (IPS), Maddala and Wu (1999) (MW, ADF) and maddala and Wu (1999) (MW, PP). These tests are applicable to a balanced panel. Indeed, LLC can be considered as a unit root test in combined panel. IPS represents a test of the heterogeneous panel as to MW is a test of unitary root in the panel, which is considerate as a non-parametric test. Table 2: Tests of Unit Root on the Variables PIB, P2, PP and RP in Panel Null: Unit root Variables LLC test Breitung t- stat IPS test MW(ADF) test MW-PP test With an Intercept PIB -0.70712 (0.2397) - 3.85971 (0.9999) 14.6645 (0.9984) 8.07640 (1.0000) Level P2-2.44171* (0.0073) - -1.82452** (0.0340) 30.0802** (0.0176) 43.0606* (0.0003) PP -3.01639* (0.0013) - -1.62062 (0.0525) 46.8796 (0.0697) 38.0789 (0.2890) RP -35.5651* - -12.8613* 77.8691 81.5615 PIB -12.3430 * - -10.6298 159.608* 187.852* First difference P2-3.32836 (0.0004)* - -3.43201 (0.0003)* 26.1701 (0.0010)* 43.2707 * PP -10.3990* -7.75995* 114.884* 116.509* RP -69.3285* - -26.8930* 160.676* 177.906* With an Intercept and Trend PIB -8.03511* 0.66434 (0.7468) -4.00805* 70.6829* (0.0002) 58.2095* (0.0060) Level P2-1.72829** (0.0420) -0.88494 (0.1881) -0.71877 (0.2361) 20.1979 (0.2114) 37.1405* (0.0020) PP -1.43220 (0.0760) 0.70214 (0.7587) 0.86209 (0.8057) 28.5237 (0.7328) 16.2500 (0.9957) RP -43.0723* 0.57152 (0.7162) -10.4207* 66.6127* (0.0003) 68.1845* (0.0002) PIB -11.9817 * -2.99677 (0.0014)* -7.23387 * 115.401 * 172.533 * 6
First difference P2-10.4722 * PP -10.1947 * -5.45539 * -5.61654 * -5.67639 * -5.71743 * 65.0129 * 90.8261 * 82.1396 * 138.484 * RP -59.1895* -3.79116* -18.7826* 148.204* 186.691* Note: *, ** represents significance at the 1% and 5% levels, respectively, of significance (bold entries). The null hypothesis is that the variable follows a unit root process, except for the Hadri Z-stat and the Heteroscedastic Consistent Z-stat. p-values are given in parentheses. Probabilities for the Fisher-type tests are computed using an asymptotic Chi-square distribution. All other tests assume asymptotic normality. The results based on the unit root LLC, IPS, MW (ADF) and MW (PP) tests with constant, and in table 4. The tests verify that all data are judged nonstationary in level. At first difference all the series are integrated meaning I (1). This unique order integration variable helps us to apply the Johansen cointegration approach panel to examine the long-term relationship between selected variables. 3.2.2 Pedroni Cointégration Tests When variables are integrated of order (1) we apply cointegration tests to examine the existence of the long-term relationship between variables. Several empirical studies are made on this level, we mention Maddalla and wu (2000) tests, Kao (1999) and Pedroni (2004) the present study, we retain the Pedroni test (2004) due to its popularity. The empirical equation cointegration tests of Pedroni is modelled as follows: PIB it = φ it + γ t i + φ 1t P2 it + φ 2t PP it + φ 3t RP it + μ it Eq 1 With i= 1,2,..n indicates each country in the panel, t= 1,2,.n and designates the period of time which can be used in the panel. The Pedroni cointegration, which includes four round statistics and three statistics of the group. There is cointegration between variables if these statistics reject the hypothesis of cointegration absence. The results of Pedroni cointegration tests are detailed in table 3: Table 3: Test of cointegration of Pedroni Methods Series: PIB, P2, PP et RP Within dimension (panel statistics) Between dimension (individuals statistics) Test Statistics Prob Test Statistics Prob Panel v-statistic -3.185556 0.9993 Group ρ-statistic 2.833472 0.9977 Pedroni Panel rho-statistic 1.128225 0.8704 Group ppstatistic -4.820763* 0.0000 7
(2004) Panel PP-Statistic - 3.237639* 0.0006 Group ADFstatistic -5.434378* 0.0000 Panel ADF- Statistic - 4.147400* 0.0000 (Weighted statistic) Panel v-statistic -0.761666 0.7769 Panel rho-statistic 1.231562 0.8909 Panel PP-Statistic - 3.585442* 0.0002 Panel ADF- Statistic - 4.134397* 0.0000 3.2.3 Granger Causality Result Because the cointegration tests are incapable of determining the sense of causality, we need a test that verifies this causality. The granger causality test is the first candidate since the linkages and causal relation stumps between variables can be examined thinks to this test. Engle and granger (1987) have shown that the causality test based on autoregression vector model (VAR) in first difference will be miss-specified when the variables are cointegrated. In this study, the economic growth measured by (PIB), the corruption measured by (P2), the oil production (PP) and the oil revenues (RP) one integrated in order(1) whether I(1), to resolve this problem, we specify an vector error correction model (VECM). To estimate a vector error correction model we increase the regression of the vector into regression model with a term period of error correction. The dynamic model of error correction seeks to model a relationship with several explicative variables and it is based on the following regressions to detect the linkages of causality in a panel data [Apergis and Pyne (2009)] Eq 2 Eq 3 k k k ln PIB ln PIB ln P2 ln PP it i1 1ip it p 1ip t p 1ip t p p 1 p 1 p 1 k ln RP ECT p 1 1ip t p 1i it 1 1it k k k ln P2 ln PIB ln P2 ln PP it i1 2ip it p 2ip t p 2ip t p p 1 p 1 p 1 k ln RP ECT p 1 2ip t p 2i it 1 2it Eq 4 8
k k k ln PP ln PIB ln P2 ln PP it i3 3ip it p 3ip t p 3ip t p p 1 p 1 p 1 k ln RP ECT p 1 3ip t p 3i it 1 3it Eq 5 k k k ln RP ln PIB ln P2 ln PP When it i4 4ip it p 4ip t p 4ip t p p 1 p 1 p 1 k ln RP ECT p 1 4ip t p 4i it 1 4it is the operator of difference, ECT is the term of error correction of cointegration long-term relationship,,,, and and are the estimation parameters? k is the lag order determined by the information criterion for testing whether granger causality from P2 to PIB. The null hypothesis ( H ) is 0 0 H : 1 0 for (i) and (p) if (h) is rejected meaning 1ip is different to zero. This hypothesis suggests that the past value (P2) has a linear. significative predictive effect on the actual value of (PIB). This hypothesis denotes that granger (P2) effects (PIB) and vice versa. Table 4: The VECM Granger Causality Analysis ip Short- run Long - run Dependent Variables PIB - Excluded variables: block homogeneity PIB P2 PP RP ECMt-1 1.631714 (0.8974) 49.55599 * 1.503371 (0.9127) 0.004807 (0.00097)* P2 3.548751 (0.6160) - 0.435924 (0.9943) 9.256675 (0.0993)*** 0.000156 (7.7E-05)* PP 5.305795 (0.3797) 1.011311 (0.9616) - 5.069668 (0.4074) -0.000253 (0.00587)* RP 11.67650 (0.0395)** 1.487553 (0.9145) 25.02391 (0.0001)* - 0.000889 (0.00041)* 9
Figure 1: Causality Relationship for Studied African Countries Oil production Oil fkfkl Gross domestic product Oil revenues Corruption Short-term Long-term In short-term, the oil production affects the gross domestic product and the oil production is on the threshold of 1%. The oil revenues affect the production and the corruption is on threshold of 10%. The gross domestic product affects the oil revenues on the threshold of 5%. All the results are approved in long term. 4. Discussion We have found that the oil revenues affect the corruption significantly which leads us to associate generally the high values of corruption for the natural resources by not omit that are many other variables that affect the couple corruption and natural resources which we may consider when we want to go further in our study. This result is confirmed by Bhattacharyya and Holder 2009 who shows that the natural resources feed the corruption. This has been verified in the case of countries with weak democratic institutions (meaning low index of corruption). There is equally an interaction relationship between the oil production and the oil revenue. Indeed, countries which are rich in natural resources make considerable profits which increase the corruption. The decision of resources exploitation is up to well-placed officials in the government hierarchy that may prompt them to resort to corruption (the higher the amount of oil produced is, the more gigantic the revenues are, the higher the corruption is). This approach was validated by Ross (1999 b). Who stated that one of resources curse is the high dependence on the resources. The analysis shows the absence of all direct link between the PIB and the corruption but this doesn't mean that the contrast is false. The corruption is made with weak efficacities of bureaucracies and bad decisions, which contradict good governance. 10
Papyrakis and Gerlagh (2004) enhance the corruption canal. Indeed Kronenberg (2004) realised an analysis on the west old countries that bureaucracies corruption reflects bad economic performances of concerned countries without being connected to natural resources but the soviets heritage plays a role as well. This leads us to explicate the absence of relationship by rejecting some factors that reveal the direct effect of the corruption on the PIB, for example, the poverty index or the technical progress. 5. Conclusion We have analysed a complicated mechanism by which the natural resources root the corruption. By using an error correction model estimation we have shown that the oil revenues has an impact on the institutions and governance. Indeed, many oil-exporting countries suffer from serious problems of corruption, which hinder any development policy or the fight against poverty. Beside results of works done by a number of authors, on modelling leads to the same results. Our analysis is focusing on 17 countries that produce oil and from the African continent. These countries suffer from civil rights deterioration and consequently the emergence of autocratic systems because of the oil richness that they have. The importance of financial flow generated by the oil industry creates equally rent-seeking behaviours by the political Elite and the economic entrepreneurial. Numerous searches modelled these activities (rent-seeking models). They all end with the same results. The natural resources sector expansion leads to low total revenue and low benefit of the economy. Deficits in governance in the rentier states have even led to extreme cases of institutional collapse. Oil resources have sometimes generated serious situations of conflict and civil wars around the ownership of the oil money. We interrogate if it is the case of all countries rich in natural resources? A future study of a different sample to ours can reveal important results. In addition, the study of governance indicators confirms that the institutions of oil countries have inferior quality than those which have the same level of development and vice versa. This can be explained by the variable level of corruption and the quality of institutional structure mentioning: regulation, laws respect, politic right, government efficacity and civil rights. The question does the poor performance of development and economic growth is allocated to governance deficits or other economic mechanisms? References Aslaksen, S., 2007, Corruption, and oil: evidence from panel data Unpublished Manuscript. Ades, A., Di Tella, R., 1999. Rents, competition, and corruption. American Economic Review 89, 982 993. Apergis, N and Payne, E. J., 2009. Energy consumption and economic growth in Central America: Evidence from a panel cointegration and error correction model. Energy Economics 31, 211 216. Bhattacharyya, S., Holder, R., 2009. Natural resources, democracy, and corruption, OxCarre Research Paper 2009-20. 11
Collier, P., Hoeffler, A., 2009. Testing the neocon agenda: democracy in resource-rich societies. European Economic Review 53, 293 308. Engle, C.W.J. Granger, 1987 Cointegration and error correction representation: Estimation and testing, Econometrica, 55, 251 276. Kronenberg, 2004. The curse of natural resources in the transition economies, The Economics of Transition, The European Bank for Reconstruction and Development, vol. 12(3), pages 399-426, 09. Haber, S., Menaldo, V., 2011. Do natural resources fuel authoritarianism? A reappraisal of the resource curse. American Political Science Review 105 (1), 1 26. Papyrakis, Elissaios and Gerlagh, Reyer, 2009, Natural Resources, Innovation, and Growth, Journal of Comparative Economics 32, 181 193 Keefer, P., 2007. Clientelism credibility and the policy choices of young democracies. American Journal of Political Science 51, 804 821. Knack, S., Keefer, P., 1995. Institutions and economic performance: cross-country tests using alternative institutional measures. Economics and Politics 7, 207 227. Leite, C., Weidmann, J., 2002. Does Mother Nature corrupt? Natural resources, corruption, and economic growth. In: Abed, G., Gupta, S. (Eds.), Governance, Corruption, and Economic Performance. IMF, Washington, DC, pp. 159 196. Marshall, M., Jaggers, K., 2002. Polity IV Project: Political Regime Characteristics and Transitions, 1800 2002. University of Maryland. Mauro, P., 1995. Corruption and growth. Quarterly Journal of Economics 110, 681 712. Levin, A., Lin, C. F., & Chu, C. S. J. (2002). Unit root tests in panel data: asymptotic and finite-sample properties. Journal of econometrics, 108(1), 1-24. Pedroni, P. (2004). Panel cointegration: asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis. Econometric theory, 20(03), 597-625. Ross, M., 1999b. The political economy of the resource curse. World Politics 51, 297 322. Ross, M., 1999a. Does oil hinder democracy? World Politics 53, 325 361. Treisman, D., 2000. The causes of corruption: a cross-national study. Journal of Public Economics 76, 399 457. Tsui, K., 2010. More oil, less democracy: evidence from worldwide crude oil discoveries. Economic Journal 121 (551), 89 115. 12