3. The Effect of Corruption on Foreign Direct Investment at the Regional Level: A Positive or Negative Relationship?

Corruption has been long considered one of the greatest obstacles to economic and social development. Already in 1998, a survey of more than 150 high-ranking public officials and key members of civil society from more than 60 developing countries ranked ‘public sector corruption as the most severe impediment to development and growth in their countries’.¹ The United Nations Secretary General António Guterres, citing estimates by the World Economic Forum, declared that ‘the global cost of corruption is at least $2.6 trillion, or 5 per cent of the global gross domestic product (GDP), adding that, according to the World Bank, businesses and individuals pay more than $1 trillion in bribes every year’.² Sustainable Development Goal 16 aims to ‘substantially reduce corruption and bribery in all their forms’.³

The World Bank defines corruption simply as the ‘use of public office for private gain’,⁴ a definition shared by Vito Tanzi from the International Monetary Fund.⁵ It includes bribery and extortion, fraud and embezzlement. Although the World Bank does not acknowledge it, this narrow definition tends in practice to limit the causes of corruption to high-ranked public officers and to excesses of regulation, and generally leads to advocating the reduction of state intervention as the principal remedy to corruption. However, corruption goes beyond the misconduct of individual public officials. It involves a ‘network of politicians, organisations, private companies, and private individuals colluding to benefit from access to power, public resources, and policy-making at the expense of the public good’.⁶ This implies that the supply side of corruption counts as much as the demand side.⁷ Moreover, the corruption of public authorities casts suspicion on big private corporations, small businesses and local traders.⁸ The impact of public authorities’ corruption on private companies is differentiated because there are ‘strategic activities conducted by firms in response to corruption’.⁹ Accordingly, the definition of corruption needs to be inclusive and encompass all possible forms.¹⁰ Synthesizing the literature, Bahoo et al. (2020) define corruption as ‘an illegal activity (bribery, fraud, financial crime, abuse, falsification, favouritism, nepotism, manipulation, etc.) conducted through misuse of authority or power by public (government) or private (firms) officeholders for private gain and benefit, financial or otherwise’. This is this definition that we use in this chapter.

There are multiple causes of corruption and have been the subject of many investigations summarised in several surveys of the literature.¹¹ For instance, Dimant and Tosato¹² list: bureaucracy and inefficient administrative and political structure; civil participation/press freedom; economic freedom; economic growth; ethnic diversity; gender; globalisation; government size; government structure; government system; historical drivers; contagious effects; economic prosperity; and education as the most important causes of corruption—the last three being considered new in the literature. Bahoo et al. (2020) perform an exhaustive bibliometric and content analysis of the business academic literature (137 articles over 17 years) and conclude that common situations, characteristics of the firm, and economic and cultural factors are the main determinants of corruption. However, analysing a comprehensive list of 36 potential corruption determinants across 123 countries (covering 87 percent of the world population), Jetter and Parmeter (2018) find that economic and institutional factors play the dominant role, not culture.

The effects of corruption on the economy encompass many issues including growth, domestic and foreign direct investment (FDI), international trade, bureaucratic inefficiency, the shadow economy, poverty and social inequalities, civil and political rights, human development, the state’s legitimacy, trust in institutions, brain drain, fiscal deficit, human capital, the environment, natural resources and climate change.¹³ The impact on the environment is a growing issue.¹⁴

For all these effects of corruption, the economic literature tries to assess whether the relationship is positive or negative. As is often the case in applied economics, there is no convergence towards a positive or a negative relation for each effect. The sample of countries and the time periods are not the same: the tested equations, the source of variables and the econometric investigations differ. Teixeira and Guimarães (2015, p. 175) have shown convincingly that ‘using distinct proxies for corruption variables, as well as controlling for other types of the countries’ institutional quality, generate distinct outcomes’.

It is beyond the scope of our chapter to address all the effects of corruption on these economic and social dimensions. We restrict our investigation to the relationship between corruption and FDI. This choice is motivated by the fact that FDI plays a key role in the economic development of the host countries, in particular in developing countries.¹⁵ More precisely, we investigate the nature of the relationship between the control of corruption and the inward stock of FDI. The relation is positive when countries which are controlling corruption successfully attract more FDI. It is negative when controlling corruption reduces FDI.

Our contribution to the literature is to look at this relation on a geographical basis. This geographical approach is based on a recent result from the literature. Corruption has a significant regional dimension.¹⁶ According to Gründler and Potrafke (2019, p. 2), ‘empirical evidence shows that corruption in an individual country or region is positively correlated with corruption in neighbouring countries or regions’. This is consistent with the above definition that corruption operates within networks, and the fact that regions like Europe or Asia have intense trade and investment internal relations prompted by well-established regional value chains.¹⁷ This geographical approach is of great interest in some regions like East Asia or Southeast Asia that have received in the recent past a large share of FDI, although they have performed poorly in terms of corruption—a phenomenon that Wedeman (2003) called two decades ago ‘the East Asian paradox’.

This paradox is illustrated by Figs. 3.1, 3.2 and 3.3. Figure 3.1 shows the evolution of the control of the corruption¹⁸ of: East, Southeast and South Asia; Australia; New Zealand; plus Europe and the EU.¹⁹ One can see that Australia and New Zealand enjoy a high level of control of corruption (+2), two times higher than Europe and the EU average (+1), which have a positive but declining score. East Asia, which regroups the most developed countries in Asia, has a low but positive score. China is among them and has gradually improved its control of corruption.²⁰ Southeast and South Asia have negative scores, which means that the perceived corruption in these sub-regions is still currently high and significant, although there has been improvement since 2006 for Southeast Asia and since 2011 for South Asia.

A multi-line graph of control of corruption versus years from 1995 to 2021. The curves are labeled Southeast Asia, Europe 30, South Asia denoting a low (2005, negative 0.3), E U 23, East Asia, and Australia-New Zealand denoting a high (2004, 2.2). The values are approximate.

In terms of stocks of FDI, although Europe and the Americas still dominate by far, Asia has increased its global share significantly since 2000, mostly at the expense of the Americas (see Fig. 3.2). A focus on East, Southeast and South Asia shows that the stocks of FDI have dramatically increased, leading to an acceleration of growth and a catch-up with Western countries (see Fig. 3.3). The progress has been much more limited for South Asia.

A multi-line graph of the percentage of world total versus years from 2000 to 2021. The curves represent Africa, the Americas, Asia, Europe, and Oceania. Europe denotes a high at (2008, 48), and Africa and Oceania denote a low at 2. The values are approximate.

A multi-line graph of U.S dollars billions versus years 2000 to 2021. Three curves represent East Asia, Southeast Asia, and South Asia. East Asia denotes a high at (2021, 4700), and South Asia low (2000, 0). The values are approximate.

Looking at the effect of the control of corruption on the FDI in 43 European countries and 39 Asian countries²¹ for the period 1996–2013, Abotsi (2018) finds that the relationship is negative for both Europe and Asia, although both regions were large recipients of FDI over the period. This confirms the existence of a paradox for both regions at the time. However, the recent evolution of the control of corruption and FDI in Asia leads one to wonder if the ‘East Asian paradox’ still exists.

To answer this question, we establish robust regional effects between corruption and FDI. Our contribution to the literature is threefold. First, we explore a large macroeconomic panel of 163 countries to provide a more reliable inference on the relationship. Second, we use a country-fixed-effects model to consider cross-sectional differences between countries in an era of financial integration, the 2002–2019 period. Third, we focus on the regional and sub-regional difference in the relation thanks to interaction terms with regional dummies.

The rest of this chapter is organised as follows. Section 3.2 reviews the literature dedicated to the effect of corruption on FDI. Section 3.3 presents our model and estimations of the relationship between corruption and FDI at world and regional levels. Section 3.4 discusses the results and concludes.

The relationship between corruption and FDI has been extensively examined by a large number of economists with the objective of determining if corruption has a negative effect on—or at least a correlation with—FDI or not (see also Chap. 2 in this volume for an extensive review). The substantive arguments used by most economists regarding corruption are not based on ethics or moral criteria but rather on efficiency. Does corruption help the market economy to work better, or does it create inefficiencies? This narrow view explains why in the 1960s some writers²² defended the argument that corruption can help promote development by getting around inefficiencies in bureaucracy. The importance given to bureaucracy is justified not only by those who think that the state is always the source of the problem but also because the early studies on corruption focused on developing countries, where it was believed that the state was more inefficient and more prone to corruption or cronyism. Despite the central role given to bureaucracy, not all economic studies on corruption actually check whether it is really the main cause of corruption, with a few exceptions. For instance, Mungiu-Pippidi and Fazekas (2020) find a negative effect of the administrative burden on corruption. The empirical studies on the relationship between FDI and corruption have latterly embraced a growing number of variables, taking into account various macro-economic and socio-institutional variables such as those we have listed above.

It appears that empirically, at the macro-level, the majority of studies find evidence that corruption does have an adverse effect on growth and development, and this effect is larger among lower-income countries.²³ Very few have found a positive effect of corruption on FDI.²⁴ This growing literature on the adverse effect of corruption explains why many international organisations decided to promote new international agreements to better coordinate the fight against corruption, such as the United Nations Convention Against Corruption (open for signature in 2003, and very widely ratified) and the 1997 OECD Convention on Combating Bribery of Foreign Officials in International Business Transaction (see Chap. 1 as well as Chaps. 4 and 5 in this volume). However, looking back, these broad treaties have been found not to have been very effective, as measured for example by survey and case study evidence. One of the reasons may be that corruption operates also at a regional level, so a regional coordination of the fight against corruption may be more effective. This is an additional reason for investigating the relation between corruption and FDI at the regional level.

Our objective is to obtain robust estimates of the relationship between corruption and FDI. To that end, we adopt the approach of Egger and Winner (2005) but for a much larger sample. Whereas Egger and Winner used a sample of 73 developed and less developed countries and the period 1995–1999, our sample includes 163 countries over the period 2002–2019 and 2752 observations. Our specification is the following:

F_it denotes country i’s FDI inward stock in year t as a percentage of GDP (source: UNCTAD database).

\({G}_{i\left(t-1\right)}\) is the real GDP in constant 2015 USD (source: World Development Indicators, WDI, from the World Bank). It is a proxy for foreign market size. We expect the positive impact of an increase in market size on inward FDI.

\({S}_{i\left(t-1\right)}\) is the secondary school gross enrolment share (source: WDI) and is a proxy for the high-skilled to low-skilled labour ratio. A low ratio reflects an abundance of low-skilled labour that attracts vertical FDI, and a high ratio of skilled labour attracts more horizontal FDI.²⁵ We expect a negative sign in the case of vertical FDI and a positive sign when horizontal FDI predominates.²⁶

\({G}_{i\left(t-1\right)}\cdot {S}_{i\left(t-1\right)}\) is an interaction variable between the size of the market and the secondary school gross enrolment share. It is an additional variable that also captures the importance of vertical FDI. It is a proxy for absolute differences of the skilled ratio in different countries.²⁷ Typically, a large developing country with an abundance of low-skilled workers will attract vertical FDI. A negative sign is expected when vertical FDI dominates.

\({C}_{it}\) = Control of Corruption CC, one of the six World Governance Indicators of the World Bank.²⁸ Contrary to Egger and Winner, we use the CC of the World Bank instead of the Corruption Perceptions Index (CPI) published by Transparency International, for two reasons. Firstly, the CC fits better with our definition of corruption as it includes corruption of both public and private agents, whereas the CPI estimates the corruption of public officials only. Secondly, there is a data break in the CPI series between the years 1996–2011 and 2012 onwards. Due to a change in methodology, the two periods cannot be merged.²⁹ The CC does not have this problem over the period 2002–2019.

\({\eta }_{i}\) = country fixed effect. This absorbs all time-invariant unobserved factors affecting F_it.

For instance, the distance from the source to the host country increases the transaction costs and therefore impacts FDI negatively.

\({u}_{it}\) is an idiosyncratic error, specific to country and time period.

We take the log of all the variables, and lagged values are used to avoid a potential problem of endogeneity.

We estimate this equation using fixed country effects. We are interested in estimating the effect of a change in the control of corruption through time. Thus, we use a fixed model in order to control for time-invariant country-specific factors. The Hausman tests indicate that the fixed effect model has to be selected. Besides, one can argue that the CC variable exhibits a limited time variation. In this case, the fixed effect model will be adapted to capture the effects of the control of corruption, which are time-invariant. The main results are presented in Table 3.1.

Columns 3 to 7 show that the size of the market (real GDP) is significant, and has a positive expected sign. This implies that the stock of FDI increases with the size of the host country’s market. The secondary school enrolment and the interaction term are significant, and have respectively a positive and negative sign. This means that multinational firms combine horizontal and vertical FDI at the world and regional levels. Our results differ from Egger and Winner’s findings (2005), which could not identify a significant interaction term and therefore any significant evidence of the dominance of vertical FDI. This can be explained by the emergence of several developing countries in the more recent period that have now shifted to higher income levels, and which have now their own multinational firms investing abroad. Multinational firms from Malaysia, Indonesia, Thailand and the Philippines are now investing in Southeast Asia and beyond.³⁰ Chinese multinational enterprises are doing the same.³¹

The control of corruption ‘cc’ is negative for all specifications. This means that at the world level, an increase in the control of corruption decreases the stock of FDI. In other words, an increase in corruption increases the stock of FDI in line with the ‘helping hand’ argument.³² We therefore have confirmed the results obtained by Egger and Winner,³³ but for a larger sample and a longer and more recent period. This implies that, at the world level, multinational firms accept paying bribes to bypass excessive bureaucracy, obtain access to the host country's market and resources or publicly funded projects, and circumvent laws and regulations. This means that despite the cost that corruption adds to their investment projects, they are still considered profitable. It also shows that despite the fact that 189 countries have signed the UNCAC since 2003 (see Chap. 1), paying bribes to invest in a host country is still pervasive at the world level.

However, results for the different regions can be seen in columns 5 to 7, which include dummy variables for each region. Column 5 includes a dummy variable for Europe and South Asia, which turns out to be non-significant. When we substitute Europe with the largest European Union economies (EU 15) in column 7, we can see that the dummy variables for East Asia, Southeast Asia, the EU 15, and the Pacific Island countries of Australia and New Zealand are significant. It is positive for all regions except for the EU 15, where it is negative.

Using column 7, we estimate the final effect by the addition of the estimate of cc and each regional dummy variable. The results are presented in Table 3.2. We can see that in East Asia and Southeast Asia, the control of corruption has a positive relationship with the inward stock of FDI. In other words, corruption has a negative relationship with FDI in line with the grabbing hand representation. This surprising result is consistent with Fig. 3.3, which shows that the control of corruption is positive and improving in East Asia, and negative but improving in Southeast Asia.³⁴ It tends to decline in Europe and in the EU but remains at a high level in Australia and New Zealand. The result for Europe and notably for the EU is surprising due to the existence of the 1999 Council of Europe Criminal Law Convention and the 1999 Council of Europe Civil Law Convention on Corruption (see Chaps. 1 and 5 in this volume). For Australia and New Zealand, the positive relationship between the control of corruption and FDI is in line with their reputation as ‘clean’ countries. Additionally, Canare (2017) finds that a country which is not corrupt and has stayed that way for many years tends to receive more FDI inflows. That is exactly the case of Australia and New Zealand in our sample.

Our discussion has focused on Asian countries. The results validate the interest in a regional approach. They also answer the question we raised in Sect. 3.1. There is no longer an ‘East Asian paradox’ because both East Asia and Southeast Asia now attract more FDI because they have increased their control of corruption. Our results are in line with Canare (2017) who also relies on the control of corruption to estimate its effect on a sample of 46 countries in Asia and the Pacific for the years 2006–2013. For the whole sample, Canare finds that both the level and the change in the control of corruption have a positive effect on FDI inflows. Countries with less corruption or that improve their control of corruption receive more FDI. However, he found no significant relationship when the analysis was limited to low- and middle-income countries. This means that other factors might play a role in determining FDI in these countries.

Looking specifically at Southeast Asian countries³⁵ for the same period as our study (2002–2019), Nguyen et al. (2021) conclude that FDI inflows are positively affected by GDP growth and trade openness. In addition, control of corruption has a positive effect on FDI inflows. Although they used a different statistical approach,³⁶ these results confirm our own findings. In another study dedicated to that sub-region, Lustrilanang et al. (2023) reach the same conclusions.

However, our findings do not confirm those of Khalid (in Chap. 2 in this volume) who investigated a sample of 33 Asian countries for the period 1995–2022 and found evidence instead of a ‘helping hand’ effect. The difference is probably explained by the difference of sample and methodology. Khalid retains the United Nations broader definition of ‘Asia’, which includes Western and Central Asia, whereas our investigation is focused on East, Southeast and South Asia. The difference of geographical areas is significant because certain variables affect countries differently, and the interplay between macroeconomic variables and institutions is not the same in area or another.³⁷ This is all the more the case given that corruption has spillover effects in neighbouring countries in Asia.³⁸ That explains why for East and Southeast Asia, we found instead a ‘grabbing hand’ effect. The choice of different indicators for corruption and the use of regional dummies instead of sub-samples, as in Khalid’s contribution, also play a role. For these reasons we believe that there is no contradiction but rather complementarity between the two chapters.

In conclusion, we have attempted in this chapter to investigate the relationship between the control of corruption and FDI at the world and regional levels, with a focus on the regions and countries that are primarily studied in this volume. Asia was known as one of the regions that received the most FDI in the recent past, although it was also known for its cronyism and high level of corruption. This situation was known as the ‘East Asian paradox’ because the majority of studies had concluded that a high level of corruption hampers FDI. The main conclusion of our chapter is that the situation has changed, and such a paradox no longer exists.

Our results show that Asian countries which improved their control of corruption, even if they started from a low level or had initially a high level of corruption, do attract (or at least are significantly associated with) more FDI. This is an encouraging result for policymakers, the judicial system and civil society engaged in the fight against corruption. Improving the control of corruption brings additional benefits. In a study of eight Southeast country members, Lustrilanang et al. (2023) found that a higher control of corruption leads to enhancing the quality of governance in the sample countries. This is an important result because an improvement in the quality of institutions has a positive influence on FDI³⁹ and an overall beneficial effect on society at large.