Are financial decisions of investors and the
allocation of resources affected by corruption? A
closer look to foreign direct investment in Eastern
Europe
A thesis submitted to The University of Groningen
For the degree of Master in Finance
In the Faculty of Economics and Business
2020
1
Abstract
This study provides a panel model empirical analysis of the impact of corruption on foreign direct investment inflows. Data concern countries of Eastern Europe, an area that is considered more corrupt than the rest of Europe. This thesis concerns two research methods. The first method is based on the effect of corruption, measured by Corruption Perception Index (CPI) on FDI inflows. The second uses Transparency International surveys that measure corruption of industries according to answers of market participants. Results suggest that corruption has a positive effect on FDI inflows, however when control variables are included this positive effect becomes insignificant.
2
1. Introduction
Findings from extensive research hold that the decision-making process of an investor is without question complex. According to European Investment Bank (EIB) (FDI and productivity convergence in central and eastern Europe, 2009), foreign direct
investments are able to substantially increase productivity rates in a number of sectors. Other types of flows like remittances and foreign aid are usually targeted at consumption-type expenses or aimed at sectors that could improve life expectancy and not towards sectors that can increase growth (Angelov, 2018). According to the United Nation’s Millennium Development Goals (MDGs), governments of
developing countries should continue building policy strategies that attract FDI. Research has led to some relatively robust conclusions when it comes to factors that influence foreign investors in their allocation of resources into a host-country company via foreign direct investment (FDI) and it has been obvious that there is a large variety of components that could influence their decisions. More specifically, direct investors seem to favor stable and democratic institutions (Ahlquist, 2006). Among others, important FDI determinants are also a country’s degree of openness, its initial development, and its legal origins. While certain drivers have been
established, a quite controversial subject concerns the corruption level and if investors’ behavior and choices are influenced by it. Therefore, corruption effects remain an important obstacle not only in the investor decision making literature but also for development economics since there is evidence that FDI inflows can be beneficial to a country’s growth rates, even though results seem not be robust in certain developing countries (Lensink and Morrissey, 2006). Over the past years, extensive research has been conducted to understand the drivers of growth as a tool for poverty alleviation and FDI effects received special attention. As it was stated this year in working papers of the European Investment Bank (EIB) (Impact of FDI on economic growth, 2020), companies in middle-income countries are more likely to be able to absorb the benefits that accompany a foreign direct investment and as a result increase their productivity rates and outputs. On the other side, according to EIB uncontrolled corruption and bad quality of institutions in these countries could bring unwanted results. Since this topic is closely related to development economics, most papers targeting this highly debated issue have been focusing on areas like sub-Saharan Afrika and South East Asia, which are dominant on the development
literature due to high poverty rates. This paper will be focusing on Eastern Europe, an area, which despite its optimal for research characteristics has been relatively
excluded from the discussion. One reason for that might be the numerous within variations of this area, which range from cultural characteristics to different
3
country variations, if treated correctly, could possibly be the key to some robust conclusions on corruption effects. Apart from the problems created by local
particularities, papers that study FDI drivers often suffer from common econometric problems such as endogeneity, heteroscedasticity, omitted variables and reverse causality. According to the European Investment Bank (Wind of change: Investment in Central, Eastern and South Eastern Europe, 2017), these countries also differ in their investment needs. More specifically, the bank concluded that some countries, like Bulgaria, Czech Republic and Hungary could use FDI in the sectors of education and infrastructure. Accordingly, for some other countries, like Estonia, Lithuania and Montenegro, investments in the energy sector would be more beneficial. The bank also concluded that Eastern Europe is in great need of FDI in the private sector, sustainable development, tourism, renewable energy and transport, therefore this area, with its centrally planned economic past, could be considered a fertile for investments ground that could work as a magnet for foreign companies.
Foreign direct investors with their resources can lead to improved productive
technologies and then to an increased output, which is vital for local firms and for the countries in their effort to catch up with the rest of Europe’s economies. Even though the transition process has started many years ago many eastern European countries, which are also members of the E.U., are still considered to be middle-income countries and some of them according to the IMF are still developing countries and economies. Some of the Eastern European countries are quite similar in terms of size, income, human capital and even culturally. On the other hand, these countries have huge differences when it comes to openness in trade and institutional quality indexes. These results as mentioned above are quite important for a foreign investor or an MNC when choosing where to allocate its resources. One aspect that differentiates Eastern European countries from others is that on average their corruption level is much higher. Even though there are exceptions, according to Transparency
International, Eastern European countries score lower by approximately 15 points on average when it comes to the Corruption Perception Index (CPI) compared to the rest of Europe. Accordingly, the average corruption performance for Eastern European countries for 2018 was 55, whereas for the Netherlands the score was 18. There is evidence that corruption in Eastern Europe severely affects local firms and companies on their decision-making (Blagojević and Damijan, 2013).
4
something that is also connected to one of the most disputed topics in development finance. More specifically, I propose a paper that within a regression framework examines the following: Do publicly available data from eastern European countries prove that a country’s FDI inflows performance is affected by its corruption level? Section 2 briefly summarizes the theory, relevant literature on the topic of FDI and corruption and several hypotheses that will be tested. In section 3, I present the data sources, the variables and the models of this paper. In section 4, I present the results derived from the model and finally section 5 concludes with some observations.
2. Literature Review
This segment of the paper investigates the theory and empirics on the relationship between corruption and FDI inflows. Subsection 1 presents various definitions of corruption and contrasting perspectives when it comes to the expected effects of corruption on FDI inflows. Subsection 2 presents findings in the literature concerning the relationship between the two main subjects of this paper, corruption and FDI inflows.
2.1 Theory
The number of factors that could be taken as potential FDI drivers is considerably large. The results in the literature are far from consistent for various reasons. In order to study the relationship between corruption and FDI, a discussion on the definition of the first should be made.
This paper will treat corruption as the misuse of public power for private benefit. This is very close to what Macrae (1982) described as corruption, an arrangement or transaction between two sides, which influences the allocation of resources now or in the future and abuses public or collective responsibility for private gain.
The most common viewpoint when it comes to corruption is that it is a dishonest behavior, mainly by those in position of power like government officials or business managers. It can include under the table transactions, laundering, bribery or even manipulating elections. This is also the definition of corruption according to Transparency International. The motive behind this behavior is the private benefit, without taking into consideration the social burden. Supporters of this approach include Globerman and Shapiro (2002), Wei (2000) and Simon and Eitzen (1990). This perspective is closely connected to corruption leading to unwanted outcomes, mainly for the economy of the country and consequently the everyday life of its inhabitants. Advocates of this approach believe that corruption is costly for investors and nepotism, briberies and other similar behaviors can increase an investor’s
5
investors do not favor this environment and only see corruption as an unnecessary obstacle. Therefore, if this viewpoint is the one closer to reality, then we should expect a negative relationship between corruption and FDI inflows.
An opposite viewpoint is that corruption can have a positive relationship with FDI inflows. Some investors may see a country with high corruption as an opportunity to avoid hostile laws, regulations and ultimately achieve their goal faster and easier (Lui, 1985). This is often observed in not only low but also middle-income countries, mainly due to poor institutional quality and a slow judicial system. Eastern Europe includes a considerable number of middle-income countries and thus it would be interesting to examine that. Advocates of this approach see corruption as part of the transaction costs and claim that the benefits of doing business faster together with the other benefits that corruption could offer to an investor, far outweigh the losses created due to increased transaction expenses (Rose-Ackerman, 1978). Some studies, like those of Leff (1964) Leyes (1965) and Bayley (1966), presented some positive externalities of high corruption. Countries with heavy bureaucracy, which is considered to repel investors, have an opportunity via corruption to receive investments from those willing to work within this corrupt framework. Haggard, (1990) stated that investors might be attracted to corrupted environments to avoid labor protection laws and taxes. These studies focused on the benefits of FDI in third world countries. Even though, Eastern Europe has a much higher quality of life, Europeanisation and liberalization levels are low in some countries, hence problems like those created by heavy bureaucracy are still present.
2.2 Effects of corruption on FDI
The importance of an investor’s decision-making process led many researchers to study the contributing factors that would motivate an investor or an MNC to allocate resources to a specific project. FDI and FPI determinants are topics that always attracted the attention of economists. As concerns the case of foreign direct
investment, trade openness is commonly mentioned as crucial for high FDI rates as well as GDP per capita (Sebastian-Andrei Labes, 2015). Hermes and Lensink (2003) also argue that the development of the financial system of the recipient country is an important element for FDI. Institutional quality and political stability also seem to be favored by foreign investors (Ahlquist, 2006) together with proximity and human capital development which are also considered important factors for high FDI inflows (Hattari and Rajan, 2011). In a very influential paper, Borensztein et al. (1998)
6
More specifically, some studies have found a negative relationship between FDI and high host-country corruption, some found a positive relationship, while others did not find any significant evidence supporting a relationship between the two.
2.2.1 Negative Relationship Findings
Consistent with the view of corruption affecting negatively FDI inflows, Freckleton et al (2012) in their research on FDI, growth and the importance of corruption found that a lower corruption level can improve the impact of FDI on growth for developing countries. Consequently, their research, even though on a different sense, illustrates a negative relationship between FDI and corruption. Accordingly, Wei (2000), when examining the relationship between corruption and bilateral FDI flows in his 57-country sample and by using a variety of corruption “indexes” concludes that the latter has a negative significant effect on FDI flows. Furthermore, the author finds that the impact of corruption on FDI is much larger when compared to other forms of capital flows. Wei (1997) also concludes that the uncertainty created by corruption also has a negative effect on FDI inflows. Bailey (2018), by using a Hedges and Olkin meta-analytic procedure of 97 prior studies based on institutional factors and FDI attractiveness also finds a negative relationship between FDI inflows and corruption. Mauro (1995) finds that a high level of corruption negatively affects the volume of investments and consequently can have a negative effect on a country’s financial development. Finally, Habib & Zurawicki (2002) in their 89-country analysis
conclude that the impact of corruption on FDI is significantly negative and stated that foreign investors prefer not to risk their resources in a corrupt environment.
2.2.2 Positive Relationship Findings
7
2.2.3 Insignificant Results
Lastly, there is a quite substantial number of studies, which failed to find a significant corruption effect on FDI inflows. More specifically, Abed and Davoodi (2000), after using a two-step process of a cross-sectional and then a panel model analysis found initially that a host-country with low corruption level is preferred by foreign investors, since these countries attract more FDI inflows per capita. In their paper, this finding is later disputed since after using structural quality as a control variable the former results became statistically insignificant. Stein and Daude (2001), did not manage to identify a significant relationship between FDI and corruption, when examining a group of source and a group of host countries. Akçay (2006) after using a cross sectional analysis based on developing countries’ data fails to find any statistically significant results on the effects of corruption on FDI and confirms that among others, openness in trade is a crucial determinant of the decision making process of an
investor. Wheeler and Mody (1992), even though their research did not focus on corruption, such as the papers mentioned above, but on a more general topic such us investment location decisions, failed to find any significant effects of corruption on foreign investments and they concluded that agglomeration benefits are the most important determinants.
A common method with which researchers tried to study the relationship between the two is by examining bilateral FDI flows and the difference-distance between home and host country’s corruption level. Bradaa, Drabek, Mendez, and Perez (2019), found that FDI flows are negatively affected by corruption. One crucial finding in their paper is that differences in home and host country corruption levels are also especially important. Firms develop certain skills, which are dependent to their home-environment. Then they search for countries with similar conditions and corruption levels since there they will be able to utilize their own skills. In countries with
8
It is obvious that this controversial topic is still a fruitful ground for research. Many of the papers in the literature used different methods and tools in accounting for
corruption and other components of their model such as institutional quality or political stability. This paper employs data from 2003 to 2018 for 18 countries of Eastern Europe. Most papers in the literature mainly targeted third world countries by using cross sectional models. This possibly can create conclusions that suffer from time-trends. The 16-year period investigated in this paper should be enough for treating temporary trends in corruption levels and FDI inflows and thus provide a more realistic point of view for each country. The knowledge gained from previous outcomes in the literature will guide the structuring of this paper’s research methods. The former together with the almost unique characteristics that Eastern European countries, with their relatively high corruption scores, need for foreign investments and the beneficial externalities possess, might be able to shed new light to this particularly difficult topic.
To be more specific I suggest a model that targets several hypotheses that aim to confirm conclusions made in the literature but also to find answers to some of the remaining gaps by using alternative data based on industry effects on a country’s corruption level.
Below I summarize the empirical hypotheses that the models of my paper should target according to the findings and the disputes inside the literature.
Hypothesis 1 (H1)- FDI will be increasing with less corruption.
Hypothesis 2 (H2)- A higher percentage of corrupt industries will lead to lower FDI inflows.
3. The Main Model
This section presents the model formation, the data sources used for the empirical analysis and finally the variables of the paper.
3.1 Model formation
Literature does not provide a single model that has proven to include all the variables necessary to held constant in order to answer our hypothesis. In this paper, I propose control variables connected with political stability, institutional quality, human capital development and macroeconomic data. The panel model regression that describes the research proposed in this paper is the following:
FDIGit = α + β1CORRit + eit (1)
9
After the inclusion of the control variables the second regression can be written as follows:
FDIGit = α + β1CORRit + β2Μit + β3Ζit + β4Iit + μi + eit, (2)
where μi is the country fixed effect. Set of variables M is connected with political risk,
rule of law and institutional quality. Vectors Z and I are sets of variables included in the regressions, such as variables connected with financial performance and
macroeconomic data for Z and variables describing human capital development for I. The structure of the data set is an unbalanced panel. Since there is no reason to assume that there is no endogeneity present, these models usually employ panel fixed effects. By including fixed effects, I control the average differences across countries in any observable or unobservable predictors. One of the main characteristics of Eastern Europe is the various between differences among countries. To be more specific, some countries are part of the European Union, some are not. Some are in the Eurozone, while others are not. Furthermore, there are substantial differences in culture, human capital and Europeanisation levels. The fixed effect coefficients will soak up all the across-group action.
Since the number of entities-countries is not extremely large, which would demand the inclusion of a large number of dummies, we can easily use both the LSDV and the within estimator in order to formulate the estimator. Referring to the first hypothesis of the paper in case of significant results:
For H1: β1 > 0 means that lower corruption leads to higher FDI inflows.
3.2 Data Sources
The central hypothesis to be tested in this paper is that corruption affects foreign direct investment inflows, controlling for political stability, institutional quality, and financial performance variables. The data used for the estimations are taken from various sources. Data on the FDI inflows will be taken from the World Bank database and the UNCTAD. Data on corruption will be based on the Corruption Perceptions Index (CPI) of the Transparency International database. Information for variables connected with financial performances and human capital development will be taken from the World Bank’s Development Indicators. Data on political risk and
10
3.3 Variables Outcome variable
Concerning my outcome variable, which focuses on FDI levels, I will use a method implemented also by Lensink and Morrissey (2006) and Ahlquist (2006). This variable is given by FDI net inflows as a percent of GDP and it concerns the period 2003-2018. Data on this variable is obtained from the World Bank’s database. This series shows net inflows (new investment inflows less disinvestment) in the reporting economy from foreign investors and is divided by GDP. The name of the variable will be FDIG.
Main Explanatory variable
CORR is a variable describing corruption. For this variable, I will use one of the most famous corruption indexes. The variable will be based on the Corruption Perceptions Index (CPI) from the Transparency International database. The Corruption
Perceptions Index ranks countries and territories by their perceived levels of public sector corruption, according to experts and business people. Values have been transformed so that 100 signifies total corruption and 0 no corruption. The average country score for 2019 was 57 out of 100. The average in my sample is also 57. This is particularly surprising, since the 2019 Transparency International average also includes third world countries.
Controls
Several variables proxying political (in)stability and institutional quality are selected to augment the basic model.
I propose a variable called INQLAW. This variable is based on data from the World Governance Indicators and more specifically on the “rule of law” index. This index has also been used by Alquist et al. (2018). Rule of Law captures the confidence levels that the members of the society have on rules such as the quality of contract enforcement, property rights, the police, the courts and the likelihood of crime and violence. Estimates give each country’s score in units of a standard normal
distribution ranging from -2.5 to 2.5 with values closer to the upper limit meaning better institutional quality.
11
The first variable will be named POLRISK and is based on data from the Political Stability and Absence of Violence/Terrorism estimate of the Worldwide Governance Indicators offered by the World Bank. This estimate measures perceptions of the likelihood of political instability and/or politically motivated violence, including terrorism. It takes values from -2.5 to 2.5 with values closer to 2.5 meaning high political stability and extremely low violence levels.
The second variable will be named VOACC. This variable is based on the Voice and Accountability estimate offered by the World Governance Indicators. The index considers freedom of expression, free media and naturally among others freedom when it comes to selecting the government. Each country’s score is given in units of a standard normal distribution, i.e. ranging from approximately -2.5 to 2.5. Values closer to the upper limit describe a more democratic environment.
A variable concerning human capital development levels for each country is also investigated in this paper. This variable will be named HUMC. For this variable data from the United Nations Development Program will be used. More specifically, this human capital measure will be based on the Human Development Index which combines information on education, health, and standard of living.
I also include a set of variables, which is meant to describe financial performance and macroeconomic data. I propose the following measures:
The next variable will be named OPENT. As it is stated in the literature, trade
openness seems to be one of the main determinants of FDI inflows. To examine that, I will make use of a variable introduced by Lensink and Morrissey (2005). This
variable is given by the sum of exports and imports of goods and services measured as a share of gross domestic product. This variable cannot take negative values, it has a world average of 60%. Data is given from the World Bank’s Development Indicators.
The next independent variable will be named GDPPC. As in Ahlquist (2006), I will use an index which describes the GDP per capita by country. More specifically, GDP per capita is the gross domestic product divided by the midyear population. All information needed is available in the World Bank database. Data is in constant 2010 U.S. dollars.
12
Another variable of this paper will be DEBGDP. This is a variable that was used also in the paper “Capital flight and political risk” by Lensink et al. (2000). Due to
availability constraints, I will use a different source. The data needed for this variable on the general government debt as a percent of GDP will be given by the IMF
database.
I also propose another variable connected to the debt levels of each country and its default threat. The name of the variable will be CRETPR. More specifically, except government debt other factors could also function as indicators of default threat. Domestic credit to private sector refers to financial resources provided to the private sector by financial corporations. The variable will represent domestic credit to the private sector as percent of GDP, as was the case in Lensink et al. (2000). The database of the World Bank offers all material needed for the variable.
Table 1: Variable Descriptive Statistics
Variable Description Source Average Min Max
FDIG FDI net inflows as a percent of GDP World Bank 5.4 -41.1 54.6 CORR Corruption Perceptions Index Transparency
international 57.0 27.0 80.0 INQLAW Inst.Quality/ Rule of Law index World Bank Governance
Ind. 4.9 -41.1 54.6
POLRISK Political Stability Indicator World Bank Governance
Ind. 0.3 -2 1,2
VOACC Voice and Accountability estimate World Bank Governance
Ind. 0.4 -1.8 1.2
OPENT Opemess to Trade World Bank
Development Ind. 113.3 56.2 190.4 GDPPC GDP divided by midyear population World Bank
Development Ind. 10.1 6.8 27.5 CRETPR Credit to the private sector World Bank
Development Ind. 48.0 0.2 100.8 HUMC Human capital index UN Development
Program 0.8 0.6 0.9
DEBGDP Governmentnt debt to GDP IMF Database 40.2 3.8 84.7 IVADM Industry Value Added as % of GDP Transparency
international 26.2 14.3 38.8 MAN Manufacturing Value Added as a %
of GDP
Transparency
international 15.3 3.7 28.6 IVAD Industry minus Manufacturing Transparency
international 11.0 5.5 20.7 The data above concern the main variables of the first part of the analysis and data are
gathered from various sources. Column 1 presents the name of the variables. Column 2 offers a short description. Column 3 presents the source of the variables. Lastly, column 4 gives
13
3.4 FDI and Industry Approach
According to Transparency International and its International Survey on corruption and bribery, which was conducted through in-depth interviews with a large number of business executives, lawyers, accountants , bankers and officials of chambers of commerce, certain industries came out as the most prone to corruption. The main question was in which sector did the interviewees expect public officials to accept or extort bribes. The findings are somewhat expected, since the results seem to confirm the anecdotal opinions of outsiders. The Construction sector seems to be the most corrupt of all since interviewees stated that when it comes to this kind of business, they most often come across corrupt behaviors such us the need for bribery. Accordingly, the second most corrupt industry relates to arms and defence. The defence world and military equipment deals and orders are often considered by outsiders to be influenced mainly by country-bilateral relationships and geopolitics. However, according to the results of the survey it seems that this industry is also troubled with briberies, which implies that the statements of the insiders confirm the anecdotal evidence that in this industry hidden agendas are prevalent. The third place in the list is owned by the energy and petroleum sector, which already has a “bad name” not necessarily due to corruption and bribes but mainly due to their pollutant behavior. This finding, which comes from business executives and other market participants, is placing this sector in an even worse position when it comes to the public image. Lastly, the fourth most corrupt business sector according to the interviewees is the industry, which includes manufacturing and mining. The survey’s results are accompanied by a remarkably interesting statement. According to Transparency International, certain industries like manufacturing and mining are less likely to use bribes compared to the industries in the first two places, but they widely use bribes in their foreign business deals. These findings could be especially useful for a paper that examines the relationship between foreign direct investment and corruption.
3.4.1 Additional Variables
The World Bank offers a variety of tools, which can help examine if certain corrupt sectors can affect the FDI inflows of a country. More specifically, I plan to use two measures. The first concerns the Industry Value added as a percent of GDP for each country. This measure might sound a bit general but it can offer the information that I need in order to examine if a large percentage of GDP that is “occupied” by corrupt industries can create a trend in FDI inflows. The “Industry value added” corresponds to ISIC divisions 10-45 and includes manufacturing (ISIC divisions 15-37). It
14
corrupt sectors according to the reports of market participants. The variable will be named IVADM.
As a secondary tool, I am planning to subtract the manufacturing element, meaning ISIC divisions 15-37 from the total of 10-45 divisions. The reason behind that is that manufacturing as expected is relatively dominant on this measure and because it is also less corrupt than the rest of the industries included on “Industry Value added” estimate, it could possibly provide a flawed conclusion. The variable will be named IVAD. Therefore, I first propose the use of the total number of divisions and then the use of divisions 10-14 and 38-45 only. Simultaneously, I will analyse the
manufacturing industry alone by using the variable MAN. If corruption affects the possibility of a foreign direct investment one would expect a large percentage of corrupt industries as a percent of GDP to create a different reaction when it comes to FDI inflows compared to a case of small percentage.
3.4.2 Methodology
The data will be used in a regression framework as explanatory variables where FDI inflows will represent the dependent variable. As a second step, the variable based on the CPI will be included in the new model. Since the findings on industry level corruption retained from Transparency International do not concern country level figures, it would be interesting to examine the relationship between this variable and a corruption variable at the country level.
This leads to the adjusted model written below.
FDIGit = α + βxit + γzi t + μi + eit (3)
Where μi is the country fixed effect. Factor x represents the corruption index. Higher
values of x amount to more corruption. Accordingly, z represents the percentage of GDP coming from corrupt industries as described by Transparency International. More specifically, z represents variables IVADM, IVAD and MAN. To connect this model with the hypotheses mentioned above in case of significant results we would make the following conclusion:
For H2: a γ<0 means that more GDP coming from corrupt industries led to less FDI inflows
15
My research structure is based on the idea that the empirical literature does not offer a consistent theoretical framework for guiding my empirical work so that no single model exists that completely specifies the variables that may be held constant in order to investigate the impact of corruption on FDI. Therefore, all possible regressions that can be specified by adding any combination of right-hand side and control variables must be estimated.
5. Empirical results
In the empirical analysis, my goal is to test the theory mentioned in previous sections, findings in the literature and shed new light on the relationship between FDI inflows and corruption by taking a different approach in the second part of the analysis. Specifically, I analyse whether a country’s corruption level is partly responsible for the variation of FDI inflows across Eastern Europe. Then I include a number of control variables, some of which often have been found to be significant in the literature, in order to test the robustness of the results. Lastly, I test whether a country’s GDP highly depended on corrupt industries as described by Transparency International can significantly affect the FDI inflows for this country. The baseline model amounts to a sixteen-year period fixed effects panel data model for 18 eastern European countries
16
Table 2: Initial Regression
(1) VARIABLES ln_FDI ln_CORR 1.782*** (0.413) Constant -5.756*** (1.661)
Fixed Effects Yes
Observations 268 Number of Countries 18 Within R-squared Between R-squared 0.070 0.002 Notes: Fixed Effects are employed. ln_CORR
refers to the log of the CP index. Robust Standard errors in parentheses***significant
at 1% level, **significant at 5% level, *significant at 10%
It has been made clear in the literature that FDI drivers are quite complex, diverse and sometimes called into question. The calculations presented in table 2 possibly suffer from omitted variables bias. Apart from the literature insights, we can also use table 2 to understand how well the variation of FDI inflows is described by this model. We observe a relatively low within and between R-squared results, which can also be indicators of some missing links in the model. To examine that, the control variables introduced above should be included in the model.
17
Table 3: Institutional and political controls
(1) (2) (3) (4)
VARIABLES ln_FDIG ln_FDIG ln_FDIG ln_FDIG
ln_CORR 1.550** 1.876*** 1.902*** 2.065** (0.659) (0.461) (0.442) (0.765) ln_INQLAW 0.286** 0.168 (0.106) (0.113) ln_VOACC 0.155 0.166 (0.108) (0.195) ln_POLRISK -0.062* -0.141* (0.032) (0.065) Constant -5.155* -5.857*** -6.178*** -6.908** (2.551) (1.832) (1.745) (2.918)
Fixed Effects Yes Yes Yes Yes
Observations 241 219 206 176 Within R-squared Between R-squared 0.121 0.016 0.090 0.009 0.085 0.003 0.105 0.049 Number of Countries 17 17 18 14
Notes: Fixed Effects are employed. CORR refers to the CP index. INQLAW refers to the trust of members of the society on laws. VOACC refers to the
Voice and accountability of citizens. POLRISK refers to political stability and Absence of violence. Robust standard errors in parentheses***significant at 1 per cent level, **significant at 5 per cent level, *significant at 10 per cent
18
variables included in the model on the relationship between the FDI inflows and corruption, we observe that CORR consistently remained significant.
Since the addition of political and institutional control variables led to small
variations, it would be wise to continue adding more controls in order to examine the robustness of the results. This time, proxies connected with financial performance will be included in the model in an effort to understand if the variation observed in the depended variable, FDIG, is caused by a country’s corruption level or some other factors. Results of the model containing the financial performance control variables are given in table 4. As before, country fixed effects have been implemented.
In the first column of table 4, I include the variable representing openness to trade. For the variable OPENT, I find an insignificant result. The second financial variable, representing GDP per capita, is introduced in the second column and it yields an insignificant coefficient. The third variable connected with financial performance represents credit to the private sector. This variable, CRETPR, is also insignificant. Finally, the fourth financial variable, representing the debt to GDP ratio, yields a significant coefficient. More specifically, I find a significant at the 1% level
coefficient. For a given country over time, a 1 percent increase for this explanatory variable would lead to a decrease of 0.36 percent for FDI inflows.
In the literature, many researchers included human capital proxies in their models, with some of them finding significant results for their variables. Column 5 of Table 4 offers a variation of my model with a human capital proxy, based on the United Nations development program data, being included. This proxy is introduced at this stage and not in table 3 to secure as many observations as possible. If HUMC was included in table 3 the model would suffer from problems created by the small number of observations. The human capital proxy appears to be insignificant. One crucial observation is that when the variables are included one by one, it does not matter if the controls are significant or insignificant, since the main explanatory variable, meaning the variable representing corruption, consistently remains significant.
19
percent increase in this explanatory variable, the dependent variable should decrease by 0.64 percent.
Table 4: Financial performance controls
(1) (2) (3) (4) (5) (6)
VARIABLES ln_FDIG ln_FDIG ln_FDIG ln_FDIG ln_FDIG ln_FDIG ln_CORR 1.942*** 1.809*** 1.869*** 1.343*** 1.653** 0.543 (0.511) (0.597) (0.559) (0.428) (0.760) (0.853) ln_OPENT 0.267 1.223** (0.521) (0.563) ln_GDPPC 0.017 -0.053 (0.270) (0.461) ln_CRETPR -0.216 -0.356 (0.258) (0.327) ln_DEBGDP -0.366*** -0.644** (0.117) (0.246) ln_HUMC -0.685 -7.549 (4.148) (7.193) Constant -7.651* -6.016 -5.223* -2.707 -5.397** -3.969 (4.000) (4.472) (2.780) (1.893) (2.286) (5.211)
Fixed Effects Yes Yes Yes Yes Yes Yes
Observations 268 268 220 268 268 220 Within R-squared Between R-squared 0.072 0.003 0.070 0.002 0.086 0.005 0.106 0.000 0.071 0.004 0.198 0.011 Number of Countries 18 18 18 18 18 18
Notes: Fixed Effects are employed. CORR refers to the CP Index. OPENT refers to openness to trade. GDPPC refers to GDP per capita. DEBGDP refers to government to GDP. CRETPR refers to credit to private sector. HUMC refers to human capital index. Robust standard errors
in parentheses
***significant at 1 percent level, ** significant at 5 per cent level, * significant at 10 per cent level.
The addition of a complete model, including all controls introduced above would not be beneficial here, since the observations’ drop would be too big.
Consequently, the addition of diverse control variables in my baseline model completely altered the conclusion that could have been made from the initial
regression process, which showed that a country’s corruption level is a significantly important factor for FDI inflows. After the inclusion of the controls, corruption became insignificant. At the same time, variables connected to a country’s financial performance together with variables related to political risk and the rule of law index were consistently significant. Therefore, those factors can be described as FDI drivers.
20
begin with, in table 5 the “corrupt industries” as described by interviewees are regressed with the corruption index, meaning variable CORR. In column 1, we observe that the industries including manufacturing are significant at the 1% level. More specifically, for a given country over time if the percentage of those industries increases by 1% then the corruption index will go up by 0.56%, ceteris paribus. In column 2, the manufacturing industry alone is regressed with the index. Here, we have again a significant at the 5% level result. The beta moves towards the same direction with the index. For a one percent increase in MAN, the dependent variable will increase by 0.23%. Lastly, for column 3, where “corrupt industries” with
manufacturing excluded are regressed with the index, we observe a significant at the 10% result. As concerns, the R-squared result we observe a substantial variation between the 3 columns. The highest R-squared result is given when all industries considered corrupt are included, meaning column 1.
Table 5: Industry and corruption
(1) (2) (3)
VARIABLES ln_CORR ln_CORR ln_CORR
ln_IVADM 0.563*** (0.180) ln_MAN 0.238** (0.101) ln_IVAD 0.155* (0.088) Constant 2.188*** 3.383*** 3.646*** (0.585) (0.268) (0.208)
Fixed Effects Yes Yes Yes
Observations 287 270 270 Within R-squared Between R-squared 0.130 0.028 0.061 0.094 0.035 0.089 Number of Countries 18 17 17
Notes: Fixed Effects are employed. IVADM refers to the industry value added including manufacturing sector as a percent of GDP . MAN refers to the manufacturing sector as a percent of GDP. IVAD refers to the
Industry value added minus the manufacturing sector as a percent of GDP. Robust standard errors in parentheses***significant at 1 per cent level, **significant at 5 per cent level, *significant at 10 per cent
21
inflows and corruption? Table 6 presents the results of the three industry variables when regressed with the main dependent variable of this paper, FDI inflows divided by GDP.
Table 6: Industry on FDI
(1) (2) (3)
VARIABLES ln_FDIG ln_FDIG ln_FDIG
ln_IVADM 2.162* (1.151) ln_MAN 1.507** (0.577) ln_IVAD 0.482 (0.582) Constant -5.628 -2.604 0.226 (3.743) (1.524) (1.388)
Fixed Effects Yes Yes Yes
Observations 268 251 251 Within R-squared Between R-squared 0.040 0.240 0.038 0.265 0.008 0.000 Number of Countries 18 17 17
Notes: Fixed Effects are used. IVADM refers to the industry value added including manufacturing sector as a % of GDP . MAN refers to the manufacturing sector as % of GDP. IVAD refers to the
Industry value added minus the manufacturing sector as a % of GDP. Robust standard errors in parentheses. ***significant at 1 per cent level, **significant at 5 per cent level, *significant at 10 per cent
22
before, according to Transparency International, the manufacturing sector is especially corrupt when it comes to foreign deals.
One somewhat unexpected finding is that my results for both parts of the analysis always contain positive coefficients. If corruption affected negatively FDI inflows, we should observe negative coefficients. As mentioned in theory above, one “school of thought” is that some types of foreign investors prefer corrupt environments since through briberies and other similar activities they can cut through bureaucracy, avoid punishments, and ultimately achieve their goals faster. Findings of Table 6 are in line with the first part of the analysis, which also contained positive coefficients. The larger the share of corrupt industries in a country’s economy the more the FDI inflows it receives. However, it would be naive to immediately conclude that the coefficients of Table 6 are positive just because investors look for corrupt industries and corrupt countries. One other reason could be for example that FDI is by “nature” attracted more by industries like manufacturing. Therefore, a country with a large
manufacturing sector would attract more FDI inflows than a country with a large tourism sector independently of the corruption level of the two countries.
5. Conclusion
Motivated by an ongoing discussion concerning the importance of Foreign Direct Investments on countries' economies, their productivity rates and their final output, this thesis examined the correlation between a country’s corruption level and the amount of FDI inflows from 2003 to 2018 in Eastern Europe.
23
For the second part of the analysis, I found a significant relationship between corrupt industries and the Corruption Perceptions Index. As concerns the effects of corrupt industries on FDI inflows, the model showed that the larger the share of a corrupt industry in a country’s economy the more FDI the country would attract. This finding, if robust, would also be consistent with the second school of thought presented in literature review concerning positive effects of corruption on FDI.
My models always employed fixed effects, which assisted my study in case any of my regressions suffered from endogeneity bias. Another econometric problem that my study came across and therefore treated was heteroscedasticity. One of the main issues that papers examining FDI drivers face is the omitted variable bias. This thesis combined findings in the literature together with some new proxies in order to
minimize as much as possible the possibility of omitted variables. One problem that this paper did not inspect is reverse causality. Further research on this problem should be made in the future.
I conclude that this paper did not find any significant effects of corruption on FDI inflows in the countries of Eastern Europe, consistent with Stein and Daude (2001) and Abed and Davoodi (2000). According to the results of the first part of my paper, a possible future drop in corruption levels for Eastern European countries, possibly resulting from more Europeanization and more integration to European laws, would not lead to increased FDI inflows. This finding is quite important, since as mentioned above governments of Eastern European countries with the cooperation of the
24
Appendix
1. Financial variables correlation matrix OPENT GDPPC DEBGD P CRETPR OPENT 1 GDPPC 0.555553 1 DEBGD P 0.005248 0.074054 1 CRETPR 0.284056 0.392752 -0.10722 1
2. Political stability and institutional quality variables correlation matrix INQLA W INQCNT POLRIS K VOACC INQLA W 1 INQCNT 0.090182 1 POLRIS K 0.143999 0.736342 1 VOACC 0.152424 0.770848 0.577549 1
3. Countries included in the research
25
4. Skewness/Kurtosis tests for Normality
Variable Prob>chi 2 FDIG 0.000 CORR 0.003 INQLAW 0.000 INQCNT 0.009 POLRISK 0.000 VOACC 0.000 OPENT 0.000 GDPPC 0.000 DEBGDP 0.004 CRETPR 0.021 HUMC 0.005
5. Heteroscedasticity test for the model including all controls
Modified Wald test for groupwise heteroscedasticity in fixed effect regression model.
H0: σ(i)2 = σ2 for all i chi2 (18) = 9836.49 Prob>chi2 = 0.0000
6. Sample average of Corruption Perceptions Index for years: 2003 -2018
26
7. Table 3 replicated for minimum observations (=176)
(1) (2) (3) (4)
VARIABLES ln_FDIG ln_FDIG ln_FDIG ln_FDIG
ln_CORR 1.509* 1.925** 1.924*** 2.115** (0.747) (0.664) (0.608) (0.922) ln_INQLAW 0.317** 0.186 (0.125) (0.106) ln_VOACC 0.247** 0.573 (0.105) (0.700) ln_POLRISK -0.057* -0.122* (0.032) (0.071) Constant -5.076 -6.042** -6.320** -7.007* (2.903) (2.612) (2.410) (3.500) Observations 176 176 176 176 R-squared 0.118 0.082 0.070 0.105 Number of Country 14 14 14 14
27
References
Abed, G. and Davoodi, H., 2001. Corruption, structural Reforms and Economic Performance
in the Transition Economies. IMF Working Paper. WP/00/132.
Ahlquist, J., 2006. Economic Policy, Institutions, And Capital Flows: Portfolio And Direct
Investment Flows In Developing Countries. International Studies Quarterly (2006) 50, 681– 704.
Aidt, T., 2003. Economic Analysis of Corruption: A Survey. The Economics Journal, Vol.
113, pp. 632-652.
Akcay, S., 2006. Corruption and human development. The Cato Journal 26(1):29-48. Alquist, R., Berman, N., Mukherjee, R. and Tesar, L., 2018. Financial Constraints,
Institutions, and Foreign Ownership. Journal of International Economics Volume 118, May 2019, Pages 63-83.
Angelov, N., 2018. Foreign Aid vs. Foreign Direct Investment: Which Leads to Equitable
Benefits? 76th Annual MPSA Conference, International Political Economy Section, the Politics of Foreign Aid, Section 1.
Bailey, N., 2018. Exploring the Relationship between Institutional Factors and FDI
Attractiveness: A Meta-Analytic Review. International Business Review Volume 27, Issue 1, February 2018, Pages 139-148
Bardhan, P., 1997. Corruption and Development: A Review of Issues. Journal of Economic
Literature 35: 1320-1346.
Bayley, D.H., 1966. The Effects of Corruption in a Developing Nation, Western Political
Quarterly. Vol. 19. N4.
Belgibayeva, A. and Plekhanov, A., 2015. Does corruption matter for sources of foreign
direct investment? European Bank for Reconstruction and Development. Working Paper No.
176.
Bellos, S. and Subasat, T., 2012. Governance and Foreign Direct Investment: a Panel
GravityModel Approach. International Review of Applied Economics, Vol. 26, pp. 303-328.
Blagojević, S. and Damijan, J., 2013. The Impact of Corruption and Ownership on the
Performance of Firms in Central and Eastern Europe. Ideas.repec.org. Available at: <https://ideas.repec.org/a/taf/pocoec/v25y2013i2p133-158.html>
Borensztein, E., De Gregorio, J. and Lee, J., 1998. How Does Foreign Direct Investment
Affect Economic Growth? Journal of International Economics Volume 45, Issue 1, 1 June 1998, Pages 115-135.
Brada, J., Drabek, Z., Mendez, J. and Perez, M., 2019. National Levels of Corruption and
Foreign Direct Investment. Journal of Comparative Economics Volume 47, Issue 1, March 2019, Pages 31-49.
Costica, M., 2012. Characteristics of FDI Process in Central and Eastern Europe. Journal:
Procedia - Social and Behavioral Sciences v62 (2012-10-24): 388-392
Freckleton, M. and Wright, A., 2012. Economic growth, foreign direct investment, and
28
Globerman, S. and D. Shapiro, A. 2002, “Global Foreign Direct Investment Flows: The Role
of Governance Infrastructure”, World Development, Vol. 30, pp. 1899-1919.
Habib, M., Zurawicki, L., 2002 Corruption and Foreign Direct Investment. J Int Bus
Stud 33, 291–307. https://doi.org/10.1057/palgrave.jibs.8491017
Haggard, S., 1990 Institutions and growth in East Asia. St Comp Int Dev 38, 53–81
https://doi.org/10.1007/BF02686328.
Hattari, R. and Rajan, 2011. How Different Are FDI And FPI Flows? Does Distance Alter
The Composition Of Capital Flows? [Online] Ideas.repec.org. Available at: <https://ideas.repec.org/p/hkm/wpaper/092011.html.>
Hermes, N. and Lensink, R., 2003. Foreign Direct Investment, Financial Development and
Economic Growth. Journal of Development Studies 40(1):142-163.
Lensink, R., Hermes, N. and Murinde, V., 2000. Capital Flight and Political Risk. Journal of
International Money and Finance Volume 19, Issue 1, February 2000, Pages 73-92.
Lensink, R. and Morrissey, O., 2006. Foreign Direct Investment: Flows, Volatility, And The
Impact On Growth*. Review of International Economics 14(3):478-493.
Leff, N.H., 1964. Economic Development through Bureaucratic Corruption American
Behavioral Scientist 8: 8-14. Reprint in A.J. Heidenheimer, M. Johnston and V.T.
LeVine(Eds.), Political corruption: A handbook, 389-403, 1989. Oxford, Transaction Books.
Leyes, C., 1965. What is the Problem about Corruption? Journal of Modern African Studies.
Vol.3.N26
Lui, F., 1985. An equilibrium queuing model of bribery. Journal of Political Economy, Vol.
93, pp. 760-781.
Macrae, J., 1982. Underdevelopment and the Economics of Corruption: A Game Theory
Approach, World Development 10(8): 677-87.
Mauro, P., 1995. Corruption and growth, The Quarterly Journal of Economics, Vol.110 No.3,
pp.681-712
Rose-Ackerman, S., 1978. Corruption: A Study in Political Economy, N.Y.: Academic Press,
20.
Sebastian-Andrei Labes, 2015. "Fdi Determinants in Brics," CES Working Papers, Centre for
European Studies, Alexandru Ioan Cuza University, vol. 7(2), pages 296-308, August.
Simon, D. and Eitzen, D., 1990. Elite Deviance, 3rd ed. Boston etc.: Allyn and Bacon. Stein, E. and Daude, C., 2002, Institutions, Integration and the Location of Foreign Direct
Investment, OECD Global Forum on International Investment. New Horizons and Policy Challenges for Foreign Direct Investment in the 21st Century: OECD Publishing.
Wei, S., 1997, “Why is Corruption So Much More Taxing than Tax? Arbitrariness Kills”,
NBER Working Paper No.6255.
Wei, S., 2000, How Taxing is Corruption on International Investors. The Review of
Economics and Statistics, Vol. 82, pp. 1-11.
Wheeler, D. and Mody, A., 1992. International Investment Location Decisions. Journal of
29
World Bank Database, <available at http://data.worldbank.org/> European Investment Bank, 2020. Impact of FDI on economic growth.
European Investment Bank, 2017. Wind of change: Investment in Central, Eastern and South
Eastern Europe.
