Determinants of corruption: can we put all countries in the same basket?

Author:Gnimassoun, Blaise
  1. Introduction

    The literature abounds with arguments and empirical evidence on the negative effects of corruption.1 Corruption is considered among others as a factor inhibiting domestic and foreign investment (Wei, 2000; Habib and Zurawicki, 2002; Meon and Sekkat, 2005; Beekman et al., 2014, Gillanders, 2014), restricting economic growth (Mauro, 1995; Tanzi and Davoodi, 1998; Mo, 2001; Gyimah-Brempong, 2002; Aidt et al., 2008; Lisciandra and Millemaci, 2017; Cieslik and Goczek, 2018) and worsening fiscal deficits, inequality and poverty (Gyimah-Brempong, 2002; Jong-Sung and Khagram, 2005; Alesina and Angeletos, 2005; Glaeser and Saks, 2006; Apergis et al., 2010; Oto-Peralias et al., 2013). All governments, whether of developed or developing countries, hold up the fight against corruption as a priority objective of their economic policy. For developing countries, this commitment is sometimes a condition for receiving financial support from development partners. Despite these pledges, the level of corruption remains high and heterogeneous between countries.

    As treatment and healing are dependent on the diagnosis, several studies have striven to study the determinants of corruption by focusing on the historical, socio-cultural, institutional and economic factors that could explain it (see among others Treisman, 2000; Van Rijckeghem and Weder, 2001; Persson et al., 2003; Acemoglu and Verdier, 2000; Fisman and Gatti, 2002; Adsera et al. 2003; Serra, 2006; Pellegrini and Gerlagh, 2008; Fan et al. 2009; Henderson and Kuncoro, 2011; Ryvkin et al., 2017). (2) These studies often focus on certain determinants of corruption--e.g. historical determinants--and disregard other factors--e.g. economic determinants--and vice versa. In so doing, they do not address the issue as a whole and some key determinants are ignored. Some authors such as Treisman (2000) and Serra (2006) have understood this and propose a broader set of determinants of corruption with the intention of subsequently identifying the most significant determinants. The problem with these studies is that the determinants are treated indiscriminately. They do not distinguish between developed and developing countries and capture the specificities of countries simply through regional dummy variables. However, there is no reason to believe that the determinants of corruption in Africa are the same as those in Europe. (3) It is therefore important to investigate this possible heterogeneity in the determinants of corruption. To our knowledge, such an analysis has not been conducted on the determinants of corruption.

    This paper thus fills this gap with the particularity of making a specific diagnosis on the determinants of corruption according to the level of development and the region. (4) Several theoretical and empirical reasons underlie this investigation. First, regional specificities in terms of natural resource endowments and socio-cultural and historical conditions add special features to the determinants of corruption. Moreover, regional proximities can create the "phenomena of contagion"; which could explain why countries have similar characteristics. The disagreements among authors support this point of view on regional heterogeneities. For example, while some studies show that a socialist legal origin is associated with greater corruption and that an Anglo-Saxon legal origin would favor lower corruption (La Porta et al., 1999; Glaeser and Shleifer, 2002; Gerring and Thacker, 2005; Serra, 2006), other economists have their reservations or disagree with the results (Adsera et al., 2000; Brunetti and Weder, 2003; Pellegrini and Gerlagh, 2008). Second, all cross-country analyses on the determinants of corruption show that corruption decreases with the level of economic development. This result, which is the only point of consensus among the authors (see Serra, 2006), may mean that studies on the determinants of corruption should take into account the level of economic development that is not homogeneous across countries. For example, the administrative barriers or lack of computerization of administrations that can be important sources of corruption in developing countries may not be very relevant in developed countries to explain corruption. In these circumstances, can we really consider the same causes of corruption across all countries as is generally presented in cross-country empirical works? If not, what are the particularities or the most relevant determinants depending on economic development and region? The answers to these questions have important implications for targeted anti-corruption policy.

    This paper provides answers to these questions starting out from the broad literature on the potential determinants of corruption. To this end, it relies on variable selection techniques to address the issue of uncertainty regarding the model specification. These techniques are much better known in economics for addressing the issue of uncertainty in cross-country determinants of economic growth (Levine and Renelt, 1992; Sala-I-Martin, 1997; Fernandez et al., 2001; Masanjala and Papageorgiou, 2008). Specifically, we use the Bayesian Model Averaging approach (BMA), which, based on the observed data, identifies the most relevant determinants of corruption, without a priori concerning the corruption model specification.

    Our paper contributes to the literature in several ways. Firstly, as data have become more available and accurate, particularly for developing countries, studies on the determinants of corruption are more likely to be more accurate and informative. However, as it is impossible to take into account all the potential determinants of corruption through a single theoretical model, an empirical approach starting with the set of determinants and then identifying the most relevant of them seems to be the most optimal solution. Next, going further than previous papers on the determinants of corruption, we identify the determinants of corruption specific to the level of economic development and to the region. Finally, by providing a specific analysis framework to better understand the determinants of corruption according to economic development level and geographical location, this article allows governments as well as international organizations to better target the root causes of corruption in order to fight effectively against this scourge.

    Based on a set of 130 developed and developing countries, our results show that while some determinants of corruption are common to all countries, others are specific to the level of economic development and to the region. Indeed, for developed countries, corruption is mainly determined by the "willingness to delegate authority", while for developing countries in general, factors such as the quality of education and GDP per capita determine corruption. (5) Focusing on the specificity of sub-Saharan Africa, our results show that freedom of the press, the burden of regulation, linguistic fragmentation, political system, religion and political stability are the determinants of corruption peculiar to this region. In the region of East Asia and the Pacific, corruption is mainly explained by legal origin, religion, political stability and education. In Europe and Central Asia, we find as determinants of corruption freedom of information, the burden of government regulations, legal origin, religion, political stability and education. Regarding Latin America and the Caribbean, freedom of information and education were identified as determinants of corruption. Finally, in the Middle East and North Africa region, determinants of corruption are the political regime, political stability and education.

    The rest of the article is organized as follows: Section 2 discusses the potential determinants of corruption. Section 3 presents the empirical approach. Section 4 is devoted to the presentation of results and comments. Section 5 provides a robustness analysis of the results. Finally, Section 6 concludes the paper.

  2. Potential determinants of corruption

    The determinants of corruption have been extensively addressed in the literature since the early 1990s. These determinants can be broadly grouped into three main blocks: historical and cultural factors, economic factors and institutional factors. (6)

    2.1. Historical and sociocultural factors

    In this block of factors, we can group the legal system and the colonial origin, religious culture, ethno-linguistic fragmentation and education.

    The legal system. This is generally regarded as one of the leading determinants of corruption. Indeed, the cost of corruption in terms of probability of arrest, exposure and punishment depends on the efficiency of the legal system. Several authors highlight the fact that the common law system that characterizes Britain and its former colonies is more dissuasive than the civil law system which is present in continental Europe and its former colonies (La Porta et al. 1999; Treisman, 2000; Glaeser and Shleifer, 2002; Serra, 2006). For the proponents of this theory based on the historical roots of corruption, the former British colonies have a better code of public service because of the influence of the British bureaucracy. In this system, the functioning of the bureaucracy is focused on the procedural aspects of the law, which improves the ability of subordinates and judges to challenge the hierarchies in order to enforce the law, thereby reducing corruption (Treisman, 2000). However, the theory of the effectiveness of the British legal system is contested by several authors, such as Adsera et al. (2000), Brunetti and Weder (2003), Pellegrini and Gerlagh (2008). Moreover, like Treisman (2000), Serra (2006) shows that the colonial heritage of a country is an important determinant in explaining contemporary corruption. But colonial heritage is strongly linked with legal origin.

    Religious culture. Since the work of Weber (1958) and...

To continue reading