Get Complete Project Material File(s) Now! »

**Methodology**

Since economists have different opinions whether corruption greases the wheels of commerce (Leff, 1964; Huntington, 1968) or sands them (Mauro, 1995; Murphy et al., 1993; Toke, 2009) by slowing or even stopping the economic growth of a country, the research done in this paper tries to contribute to this question. The researchers on both sides share arguments for and arguments against their opinions. However since this is a complex issue, this paper hopes to solve, or at least contribute to a better understanding of the subject. As discussed in the data section, six explanatory variables are used together with one structural break dummy, whereof the dummy variable accounts for Montenegro not existing as an independent country before 2006. Using the panel regression study was the best option based on the available observations for the region in question since cross-sectional study covers fewer observations, and time series studies have some gaps in data availability throughout the years. Therefore, the variables gathered from year 2003 until 2014 provide one with all values and allow for strongly balanced panel data. The hypothesis which is being tested in this thesis, by running a panel study on 14 countries under the timeframe from 2003 to 2014, is to observe whether corruption affects economic growth of countries in Southern Europe and Balkan countries and if so, which kind of effect does it have.

Even though previous papers within the field of corruption (Toke, 2009; Barro, 1991; Mauro, 1996), often approach this question by doing cross a-sectional study that is not the course of this paper. Instead, the models in this paper are adjusted for panel data. To test our hypothesis, three linear regression panel data models with fixed effects are specified. All three models that are specified (Model 1; Model 2; Model 3), test for effect of corruption on economic growth. Panel study is chosen since it enables following dynamic changes in both cross-sectional and time series, but also it examines the short-run effects, which are interesting to test for.

When panel study is compiled, one can choose between fixed and random effects to get the regressions which are most appropriate for the study. Both regressions with random and fixed effects have been performed. Based on the results, it is possible to run Hausman test to determine which effects are more appropriate for the regressions in this study. Hausman test makes it possible to determine whether differences between countries are caused by random variances or not, for each country. In this way, Hausman test implies which of these two effects are more appropriate for the regressions (See Figure 0-1 in Appendix 1).

Besides discussion about effects, test for pure autocorrelation is made (See Table 0-1 in Appendix 1). Since results of the test show that there is in fact pure autocorrelation in the data, robust standard errors are used.

*Real GDP Per Capita Growth**it=β**0i+β**1*Corruption Perception Index+β**2*Lagged Real GDP Per Capita Growth+β**3*Structural Break Dummy+u**it *(Model 1)

*Real GDP Per Capita Growth**it=β**0i+β**1*Corruption Perception Index+β**2*Gross Fixed Capital Formation+β**3*Trade Openness+β**4*Population Growth+β**5*Lagged Real GDP Per Capita Growth+β**6*Structural Break Dummy+u**it *(Model 2)

Correlation matrix (See Table 0-2 in Appendix 1) is also computed in order to check for within-panel correlation structure. None of the variables exceed 0.570 correlation, which is correlation between Gross Fixed Capital Formation and Lagged Real GDP/Capita Growth. These results imply that all of the variables can be used when running linear regression panel models. In further sections, models consist of two simplified versions (Model1; Model 2) where potential different outcomes are going to be tested for. The first variation (Model 1) that is going to be tested is whether CPI, representing corruption in this study, regressed together with lagged real gross domestic product per capita growth affect the real gross domestic product per capita growth, without including the rest of the variables in the regression. In the second variation of the regression (Model 2), all variables except for foreign direct investment net inflow, as a percentage of GDP, are used.

Last, but not least is the model (Model 3), which includes all the variables in the regression.

*Real GDP Per Capita Growth**it=β**0i+β**1*Corruption Perception Index+β**2*Gross Fixed Capital Formation+β**3*Trade Openness+β**4*Population Growth+β**5*Lagged Real GDP Per Capita Growth+β**6*FDI Net Inflow+β**7*Structural Break Dummy+u**it *(Model 3)

In all three models (Model 1; Model 2; Model 3), the null hypothesis is that there is no effect of perceived corruption (CPI) in public sector on economic growth. In the section *Results*, outcomes of the regressions will be examined to see if the null hypothesis is to be accepted or rejected. Based on that, it is possible to estimate the effect of corruption on economic growth in the case of the Southern European countries^{8}.

**Results**

After gathering all the needed variables and choosing the right methodology, three linear regressions are run. Since panel study is done and the area of Southern Europe consists of the countries which are culturally and geographically close to each other, but also, politically and systematically different from one another, fixed effects regressions are used for testing. To control for this and to be sure that fixed effects fit better than random effects, Hausman test is performed. Since the outcome of Hausman test (Figure 0-1 Hausman) is not statistically significant, which means it is below 0.05, the null hypothesis is rejected. This means that differences in coefficients are systematic which implies that fixed effects should be used when running linear regression panel models.

The first regression (1) includes CPI and Lagged Real GDP Per Capita growth in order to see the effect of corruption on economic growth in the Southern European region. Result of this regression is that CPI is statistically significant variable at 10 percent. This regression tests for a clear effect of corruption on economic growth since CPI and Lagged Real GDP Per Capita Growth are included only. Regression (1) implies that increase in CPI by one unit, which means that level of corruption in the countries decreases^{9}, leads to a decrease in economic growth of the country (Real GDP Per Capita Growth) by 0.120 percent. This finding goes hand in hand with claims (Leff, 1964; Huntington 1968) that corruption serves as a greaser of the wheels of the economy. In regression (2), Gross Fixed Capital Formation, Trade Openness and Population Growth are included to see whether the significance and the effect of CPI on economic growth will change. It turned out not to be the case since CPI is still statistically significant at 10 percent with coefficient being negative and even increasing the negative effect on economic growth. This result indicates that one unit decrease in corruption would lead to a decrease in economic growth of the countries by 0.164 percent. Beside this effect, we can see that by including the three new variables in the regression (2), its R-squared increases from 0.328 to 0.358 which implies that the three extra variables explain more of the variation in dependent variable which is economic growth in the case of this study.

Trade Openness and Gross Fixed Capital Formation are good indicators of economic growth which can be seen from the regression (2), where both variables are statistically significant as determinants of economic growth. This goes especially to Trade Openness, which has been recently considered to play an important role in economic growth of the country (Pradhan, R. P.; Arvin M. B. & Norman N.R., 2015). In the last regression (3) run, Foreign Direct Investment Net Inflow is added, which is a good determinant of economic growth (Zhao, 2013), to see if there are any changes in relationship between corruption and economic growth of the countries in Southern Europe. The retrieved results imply that corruption is still statistically significant at 10 percent and that it positively affects the economic growth. One unit increase in CPI leads to 0.163 percent decrease in economic growth of a country implying that corruption still seems to be a greaser of the economy in Southern Europe. When it comes to R-squared, it slightly increases due to the new variable added and has a value of 0.359 or it explains 35.9 percent of the variance in dependent variable. Compared to other papers in this field, R-squared is on similar level with the ones (Swaleheen, 2011) that use Real GDP Per Capita Growth as a dependent variable. This is encouraging since it shows that the study does not stand out from other studies in negative sense and it explains as much as they do. Contrary, studies that have used real GDP per capita as a dependent variable (Toke, 2009) have much higher R-squared.

Since regressions run in this study are linear regression panel data models, results reported in Table 5-2 are the results obtained from controlling for heteroscedasticity and autocorrelation with robust standard errors. This way, the significance levels and coefficients are more robust since they are not affected by any of the two problems, which can appear when dealing with panel data.

**Discussion**

Even though the results show a significant value that corruption might act as grease to economic growth of a country, conclusion should not be made that this is in fact true. Corruption may seem harmful in some situations where it is considered as petty cash, however in the long-run almost all papers on the subject have brought forward the general idea that corruption has a negative effect for a country. It has been stated that corruption is such a difficult variable to find real data on, due to the given illegal nature and that corrupt transactions are typically cloaked or hidden from the public. Ambiguity with the results in this study which has been presented in previous part of the paper is that a positive effect of corruption is found. However, according to other articles (Aidt, 2009), even though corruption may have a positive effect on economic growth, it is most likely only in the short-run. Our result of corruption being a “grease” on economic growth are results based on panel data which can be interpreted as a study of the economy in the short-run. Previous studies such as the ones from Mauro (1996) and Barro (1993, 1995) conduct cross-sectional regressions by taking averages over the sample period, which could be interpreted as studies over the long-run. Both article and studies provide regressions with a significant negative relationship with corruption index towards economic growth and development.

1 Introduction

1.1 Background

1.2 Previous studies

1.3 Purpose and Contribution

1.4 Limitations

2 Theory

2.1 The Sanders and the Greasers

3 Data

3.1 Sample

3.2 Variables .

4 Methodology

5 Results

7 Discussion

8 Conclusion

References

GET THE COMPLETE PROJECT

The Effect of Corruption on Economic Growth Does corruption show a significance effect in the growth of an economy?