Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative

Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Resources Policy journal homepage: www.elsevier.com/locate/resourpol

Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative Caitlin C. Corrigan n University of Pittsburgh, Graduate School of Public and International Affairs, 3601 Wesley W. Posvar Hall, Pittsburgh, PA 15260, United States

art ic l e i nf o

a b s t r a c t

Article history: Received 24 March 2013 Received in revised form 5 October 2013 Accepted 7 October 2013

This article critically examines the impact, up until 2009, of the Extractive Industries Transparency Initiative (EITI). The EITI is an international policy intervention that aims to mitigate the negative effects of resource abundance by promoting the transparency of resource revenues and accountability of the governments of resource rich states. Its effectiveness can be assessed by examining two outcomes that are suggested to be negatively affected by resource abundance: economic development and quality of governance. Through a panel study, including approximately 200 countries, the influence of the EITI in these two areas is examined. Results suggest that the negative effect of resource abundance on GDP per capita, the capacity of the government to formulate and implement sound policies and the level of rule of law is mitigated in EITI countries. However, the EITI has little effect on level of democracy, political stability and corruption. The study concludes that there are some early indications that the EITI has been successful in protecting some nations from selected elements of the resource curse. This is encouraging given the relatively short time period since the founding of the EITI, however the mixed results suggest that a similar study should be repeated in 5 to 10 years when EITI policies have had enough time to fully take effect. & 2013 Elsevier Ltd. All rights reserved.

JEL classification: 013 Keywords: Natural resources Resource curse Extractive Industries Transparency Initiative (EITI) Transparency Resource governance

Introduction Natural resources have been sought after and fought over throughout history. They invite conflict due to their scarcity as they are highly valued yet unevenly distributed, given the varied geographic concentration of nature's bounty. Although this maxim is true to varying degrees depending on the particular natural resource in question, it has been frequently observed that the possession of these valuable resources often leads to negative development rather than the positive benefits that would be intuitively expected to accrue to the people and places that are so endowed. This phenomenon has come to be referred to as the “resource curse”. Arguably, the resource curse is not always inevitable, and many aspects of society beyond simple resource abundance have been examined as other possible factors in determining whether a country will experience such negative effects. Transparency and accountability within government are potentially among the key determinants of the economic, political and social consequences of natural resource abundance. With this paradigm in mind, among other campaigns, the Extractive Industries Transparency Initiative (EITI) has been established to increase transparency and accountability in resource-rich

n

Tel.: þ 1 216 200 9500. E-mail address: [email protected]

states in order to counteract the negative effects of natural resource abundance that have been observed. The EITI has had success in recruiting resource rich countries to become members. At the time of writing, 33 countries had either been designated as candidates for EITI compliance or were already compliant, 21 of which are Sub-Saharan Africa. This article asks the question: Has the EITI been effective in mitigating the “resource curse”? Effectiveness is assessed by examining two outcomes that seem to be negatively affected by resource abundance: (1) economic development, and (2) the quality of governance. The next section of the paper explains why these two outcomes are key elements of the resource curse. The concepts of transparency and accountability are examined in detail given their prominence in the goals of the EITI. An examination of the EITI's goals and the role of transparency and accountability within the initiative are discussed in the third section of the article. These factors are fundamental to the argument that EITI membership can be used as a proxy measure of willingness to reform institutions (i.e. increase transparency and accountability). As institutional quality has been found to be a determinant in whether a country is affected by the resource curse or not, membership is thought to improve institutional quality and therefore magnify the factors that are negatively influenced by the resource curse. The remainder of the article is devoted to providing an empirical study of the EITI's effectiveness in terms of economic

0301-4207/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.resourpol.2013.10.003

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

development and governance. This is accomplished using panel data from approximately 200 countries from the years 1995 to 2009. The results illustrate the expected negative effects of natural resource abundance on economic and governance outcomes but the evidence of EITI membership moderating this effect is mixed. Although results indicate that that EITI membership is associated with a significant decrease in the negative effects of resource abundance on GDP per capita, government effectiveness, regulation quality and rule of law, weaker and inconsistent improvements from the EITI were seen in control of corruption and voice and accountability. Unfortunately, the EITI membership had no significant effect on political stability/no violence. Results also vary to some extent depending on which indicator for natural resource abundance is used. Equally puzzling is that the EITI membership itself had a negative effect on most of the outcomes. These adverse findings may be the result of other factors such as negative selection of countries asked to join the EITI, a lack of time for implementation, the complicated nature of the EITI membership process and/or actual problems with the effectiveness of the EITI. The article concludes that taken as a whole, there are some early indications that the EITI has been successful in mitigating certain aspects of the resource curse. More research, however, is needed to evaluate the EITI's influence in the long run, allowing more time for benefits to take effect.

The resource curse: A literature review Natural resource effects on the economic development The debate The resource curse debate originated as a theory based mainly on the economic effects of natural resource abundance. Sachs and Warner (1997) demonstrated that countries with abundant natural resources tend to grow more slowly than countries without large quantities of natural resources. In their research, they suggested that the main reason for the negative growth effects observed is Dutch Disease. However, many questions remained as to why some countries with large quantities of resources tend to do well economically, while others, with similar natural inventories, tend not to perform as well. Complicating the debate is the secondary question of just how the natural resource sector within a country should be quantified. Indeed, depending on how the measurement debate is resolved, this quantification could determine if the resource curse actually exists. Sachs and Warner (2001) argued that looking at the concentration (intensity) of resource abundance in a country is more important than examining the gross quantity of resources per capita given that a high concentration (intensity) often results in a strong ‘crowding out effect'. As such, they opted to use natural resource exports as share of GDP. However, Alexeev and Conrad (2005) challenged this, countering that measuring natural resources in terms of the ratio of exports to GDP is flawed. They argue that more developed countries would consume more of their resources domestically, diluting the measurement of resource abundance made by Sachs and Warner (2001). Furthermore, they argue that because they are examining the effect of resources on GDP, measuring resources in term of it share in GDP is flawed. They instead measure per capita oil and mining output. They find that when adjusting the empirical measurement and methods, specifically for the overuse of the growth rates as an indicator and measuring resources as shares of GDP, there does not appear to be any resource curse for oil. Sala-i-Martin and Subramanian (2003) disaggregate the Sachs and Warner (1997) measure and differentiate between point source resource exports (fuels and metals) and agricultural and

raw materials. They find none of these exports have a direct effect on growth, but find that point resources did have a detrimental effect on growth through their effect on institutions (discussed more fully below). Isham et al. (2003) also distinguish between exports and find that point source and plantation exports negatively affected institutions, which in turn affected a country's ability to deal with price shocks, ultimately affecting prosperity. Leite and Weidmann (1999) went further and deconstructed the natural resource exports as a variable for their study looking at the connection between corruption and natural resources (however, Alexeev and Conrad (2005) still argue that all of these findings are questionable due to the use of a GDP as a control variable). The role of institutions The strength and quality of government and societal institutions are a possible explanation for why some countries succumb to the resource curse while others seem to benefit economically from their natural resources. The success of resource rich countries such as Botswana and Norway has prompted scholars to investigate the protective role institutions might play in determining whether countries fall victim to the resource curse. Mehlum et al. (2006) find that resource abundance only affects growth rates negatively when institutions are weak. In contrast to Sachs and Warner's work, countries that are over the threshold for high quality institutions, as defined by the authors, do not experience the negative impact of the resource curse. In addition, Sala-i-Martin and Subramanian (2003) find an indirect link between growth and natural resources with respect to institutional quality. They find that natural resource abundance has a negative effect on institutional quality and that institutional quality has an impact on growth. However, once institutional quality is controlled for, there is no effect due to natural resource abundance on growth. However, these findings only apply to “point source” resources such as fuels and minerals. Boschini et al. (2005) refer to the interaction between institutional quality and the type of resource as “appropriability”. Alluvial diamonds, for instance, are highly appropriable products, while most agricultural products have a low appropriability. The more appropriable a resource is, the more important the interaction with institutions. Boschini et al. conclude that “sufficient improvement in institutional quality turns resource abundance into an asset rather than a curse” (27–28). Robinson et al. (2006) look at the effect of resource booms in a society. In countries with weak institutions, where politicians may turn to clientelism and tend to discount the future because of their reduced chances of remaining in power, resources will often be extracted too quickly, resulting in a boom, which in turn renders the politician even more inefficient and distorting the economy even further. When occurring in the presence of robust institutions, however, resource booms will raise national incomes. This section of the paper has identified two main factors in determining the extent to which a country can suffer economically from the resource curse. One is the abundance of natural resources within a country and the other is the quality of its institutions. Since a country has little control over the quantity of resources with its borders, the quality of institutions is, therefore, the most critical determinant of outcomes. The next section of the paper will examine ways in which resource abundance can influence governance directly and draw attention to why building strong institutions in resource abundant countries could be a daunting but necessary task requiring outside assistance. Natural resource effects on governance It has also been suggested that a country's governance suffers directly from natural resource abundance. For instance, Collier (2007)

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

implies that if resource rents make up a substantial amount of government revenue, these can aid in undermining democracy through patronage politics and corruption and can break down the system of checks and balances that keep institutions strong. Ironically, it is the resource-rich countries that need checks and balances the most, and yet have the hardest time keeping them in place. Furthermore, in looking at oil and its effect on democracy, Ross (2001) finds evidence of a ‘rentier effect’, which suggests that resource-rich governments use low tax rates and patronage to relieve pressures for greater accountability; a ‘repression effect’, or the idea that resource wealth retards democratization by enabling governments to boost their funding for internal security; and a ‘modernization effect’, which holds that growth stimulated by the extraction of oil and/or minerals fails to bring about the social and cultural changes that tend to produce democratic government (327–328). Bhattacharyya and Hodler (2009) argue that depending on the level of democracy, resource rents tend to increase corruption, which not only decreases the effectiveness or the level of accountability of a government but can distort the economy as well. Busse and Gröning (2013) also find that natural resource exports lead to increased corruption, using an instrumental variable technique that accounts for endogeneity. Additionally, as has been suggested above, resource-rich governments receive so much revenue from rents that they have little need for taxation and are, therefore, less accountable to the tax-paying public. Such governments have lower motivation to push through development enhancing proposals or remain democratic. In other words, the effectiveness of the government to make quality policies can be undermined by large inflows of revenue generated from natural resources (Collier and Hoeffler, 2008; Iimi, 2007). Iimi (2007) finds that regulatory quality and control of corruption were of particular importance to the maintenance of quality institutions, resource management and the reversal of the resource curse. In addition, many studies have found that resource abundance has an effect on stability within a country. Collier (2002) finds that at a primary commodity dependence of 26% of GDP, the risk of civil war increases 22.5%. Moreover, when the types of resources are taken into account, the effects on duration and onset of civil war became clearer (Ross, 2002). Conflicts can undermine political stability, government control, the rule of law and faith in the government. These, in turn, hurt the economy in a multitude of ways, including through promoting opportunistic behavior (Collier, 2002). Studies show that conflicts reduce the annual GDP per capita growth by approximately 2.2% per year. Additionally, conflict can bring about human capital loss, investment loss (especially in manufacturing, construction, transportation and finance), production change, and infrastructural loss due to war damages (Collier, 1999). Transparency and accountability: A cure? At the center of the debate over the occurrence of the resource curse lies the importance of the role of institutions in promoting economic growth and maintaining high-quality governance. If institutions are not strong enough to oversee ownership, profit taking and resolution of grievances, the rapid flow of resource rents can quickly overwhelm the government's ability to exert control. Maintenance of strong quality institutions, therefore, is much more important in resource-rich countries. A recent trend in promoting transparency and accountability in order to reduce the negative effects of natural resource abundance has emerged (i.e. The Kimberly Process, Publish What You Pay, the International Council on Mining and Metals or the Extractive Industries Transparency Initiative) (Feldt, 2009). Transparency and accountability initiatives seek in general: (a) to improve the

3

processes through which actors and institutions can effectively bring governments to account and (b) to effectively contribute to better development outcomes, such as a more egalitarian distribution of wealth and better socio-economic conditions or poverty alleviation (Mejía Acosta, 2013, 93). The reasoning behind the call is that transparency opens the channels of communication and allows scrutiny over revenues gathered from resources as well as how the resources are generated and extracted. Accountability ensures that the government remains adherent to the needs of their citizens and not only to the revenue sources that keep them in power. Transparency and accountability within government is expected to mitigate some of the negative economic and quality of governance effects seen in countries with poor institutions and abundant resources by making it harder for governments to divert revenues to corruption and patronage.1 Under increased transparency and accountability, the government would, theoretically, invest more in pro-development policies, increasing the levels of effectiveness of government. Several studies have examined the question of how transparency and accountability affects institutional quality. Andreula et al. (2009) find that fiscal transparency matters for institutional issues and, furthermore, that institutional quality or governance is a determinant of fiscal transparency. These findings make clear that government transparency, especially concerning revenues, go hand in hand with the quality of institutions within a country. Additionally, Islam (2003) finds that countries with more information flows (i.e. higher transparency) also have higher quality of governance. Kolstad and Wigg (2008) examine the effects of transparency on corruption and find that it is necessary and yet not sufficient for reducing corruption. While transparency allows for access to information, one also needs to be able to act on the information presented. They see improvements in rule of law and in the accountability of how government revenues are used (expenditures) as vital to deterring the two major mechanisms in the resource curse, rent seeking behavior and patronage. This suggests that transparency without accountability is limited in its effectiveness on increasing institutional quality. Additionally, the question of why resource-rich countries might need more outside assistance to ensure transparency and accountability as compared to resource-poor countries should be considered. One reason that the former might be more prone to a lack of accountability is the magnitude of resource rents flowing to the government. It has been found that the more rents from resources increased, the lower the tax enforcement is within a country. Lower tax enforcement, in turn, lowers the demand for accountability (McGuirk, 2010; Ross, 2001). Moreover, Collier (2006) suggests that compared to non-oil exporting countries, oil exporting governments do not “spend more, they tax less” (1484–1485). Scrutiny is often a related result of citizens' tax burden, and as taxes go down, government scrutiny also decreases, lowering accountability and lessening the call for transparency. Additionally, resource revenues are harder to define within government spending, allowing for diversion to patronage and corruption. Therefore, according to Collier (2006), trust cannot be placed just in politicians but must be supplemented with checks and balances (institutions) as well. From these findings, it can be seen that countries with large inscrutable rents and weak taxing authority are under little pressure to be accountable to their citizens and transparent in financial matters. Additional support

1 Stiglizt (1999) outlines the adverse effects of non-openness, or secrecy; “secrecy serves to entrench incumbents, discourage public participation in democratic processes, and undermines the ability of the press to provide an effective check against the abuses of government.” Furthermore, secrecy weakens the quality of decision making, and through increases in corruption, has adverse economic consequences (14–15).

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

and incentives are needed to improve transparency and allow revenue flows to be scrutinized.

Building transparency and accountability through the Extractive Industries Transparency Initiative (EITI)

is true, then one could argue that there should be an improvement in governance and performance on economic indicators within a country, even as early as when they first decide to pursue EITI implementation and compared to other resource-abundant countries which did not pursue implementation. Limitations of the EITI

The EITI was first launched in 2002 by U.K Prime Minister Tony Blair (Extractive Industries Transparency Initiative, 2009a). It aims “to strengthen governance by improving transparency and accountability in the extractives sector” (Extractive Industries Transparency Initiative, 2013b), specifically by building international standards for resource-rich governments that lead to higher transparency and accountability. Proponents believe it can assist with stabilizing economic growth and facilitating reductions in poverty (Pitlik et al., 2010). Its 12 principles emphasize improved transparency within a natural resource sector (Extractive Industries Transparency Initiative, 2011). The EITI can be seen as an initiative to create transparency in order to strengthen government institutions. Pitlik et al. (2010) describe it as an “attempt to impede these practices of corruption with theft” (178). Through the disclosure of company payments and government revenues from oil, gas, and mining, corruption within governments, proponents believe, could be controlled. Accordingly, they see participation by a government in the EITI as a “signal of willingness to reform”. It follows, therefore, that by joining such an organization a county is accepting the validity of its international standards. If the country then does not accede to those standards, it puts itself at risk of loss of reputation as well as more tangible losses such as financial aid or foreign investment. Dreher and Voigt (2008) found that membership in international organizations is significantly and robustly linked with better credibility and that this credibility-enhancing effect is strongest in countries whose domestic institutions are weak. Countries that choose to be a part of the EITI, therefore, may have increased legitimacy in the international arena, and since most are developing countries, this effect could be even more significant. The countries that have chosen to join the EITI are countries with a higher share of natural resource exports, a higher amount of ethnic fractionalization and higher incidence of corruption, who are also democracies and not in OPEC (Pitlik et al., 2010). The process of the EITI validation brings together government, companies, and civil society to increase accountability and to promote participation by the whole society within the resource extraction process (Extractive Industries Transparency Initiative, 2013b). By requiring that reports critical of the government and industry practices be made widely available to the public, the EITI is promoting transparency within the governments receiving revenues, including all those involved in the process. The validation process, therefore, should ensure that the governments are working towards more accountability and transparency of practices. This increase in transparency and accountability should theoretically reduce the severity of some of the governance issues experienced in resource-rich countries and should strengthen connected institutions as well, thereby mitigating the negative economic effects more often felt by resource rich countries. If this

Resource Wealth

While the goals and ideas of the EITI appear sound in theory, there has been much discussion concerning its limitations, given the current structure. On the positive side, Caspary (2012) reports that the EITI has shown early signs of positive impact in terms of financial data being widely released on oil, gas and mining, coupled with the creation of effective multi-stakeholder structures, which have been seen as helping to build trust and collaboration. However, Global Witness (2009) cites other challenges for the EITI to be effective. First, there needs to be a stronger and more visible role for civil society. This leads to greater government legitimacy and, as such, civil society groups “must be able to play a full and free part in EITI”. Second, the validation process needs to be executed by an objective third party. Third, there must be obvious and achievable rewards for those that meet compliance. Forth, United States, Canada, Australia and Britain should implement the EITI. Lastly, the EITI will have to evolve to include disaggregate reporting of each country's revenues, company-by-company. Aguilar et al. (2011) examine the importance of implementing the EITI at the sub-national level and outlined the constraints that have so far accompanied these attempts. Ölcer (2009) argues that the minimum standards for the EITI are not sufficient. Inaccurate and inadequate information about revenues is still a problem for EITI candidate and compliant countries. The EITI needs to consider the entire value-added chain of exploitation and transformation of natural resources and not just the intermittent material payments. The initiative's voluntary nature and unenforceability also reduces its effectiveness. Other problems include the non-participation of key hydrocarbon producing countries, or civil society that is often not strong enough to take on the responsibility the EITI gives it. Aaronson (2011) argues that there are three reasons the EITI is not as effective as it could be: (1) the various partners (government, civil society and business) have different visions; (2) information and participation has not been fully given to civil society in many countries; and (3) in many countries the public and the legislators have not been made aware of the EITI. Thus, the most compelling power through which the EITI could work, public scrutiny, is not being adequately utilized. Hilson and Maconachie (2009) examine the EITI in SubSaharan Africa and conclude that there must already be strong foundations of good governance in place and a commitment to institutional change in order for it to be in any way effective at reducing corruption by means of allowing citizens to hold their governments accountable. Along those same lines, Smith et al. (2012) reviews the case of Madagascar, arguing that the EITI failed to address complex linkages and power relations between stakeholders. While civil society is being manipulated by local power structures, the mechanisms that allow participation to lead to

Improved Institutional Quality (through EITI membership)

Improvments in Economic Development and Quality of Governance

Fig. 1. EITI argument.

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

good governance will not be effective without outside intervention and support. While the study undertaken in this article will look at how EITI membership affects various governance indicators (as well as economic development), a further argument can be made that governance can affect EITI membership and success. The argument Much of the past research in this area, as mentioned above, has – paradoxically – linked natural resource wealth to lower levels of economic growth and development. Furthermore, it has been suggested that resource wealth can hinder democratization, control of corruption, the quality of policy and political stability. Improvement of institutions through accountability and transparency has been seen as a way to mitigate the negative impacts of resource wealth. This is the fundamental assumption behind the EITI. By compelling (or convincing) member countries to disclose payments, institutional quality should improve and thereby mitigate the negative “resource curse” effects. Fig. 1 outlines this proposition. Countries have little control over the amount of resources they have. However, they can implement measures that increase the quality of their institutions, which should mitigate the negative economic development and quality of governance effects of the resource wealth. If the EITI is, as argued above, a sign of countries' willingness to reform institutions and increase transparency and accountability, then membership should lead to improvements in both economic and governance related indicators. Since there is certainly no consensus on the effectiveness of the EITI, testing these arguments, empirically, would be invaluable.

Has the EITI been successful? Definitions and methodology The assumption made in the literature is that without quality institutions, an increase in natural resource exports will negatively affect GDP per capita. This section of the paper reports on an empirical analysis undertaken to determine whether EITI membership is associated with economic and governance improvements among countries with high levels of resource abundance. The research design for determining the effects of EITI membership is a pooled cross-sectional panel study (Yaffee, 2003). In this case, the cross-section is 200 countries (list modified from World Bank statistical database) and this panel is followed for a time span from 1995 to 2009. This method was used in order to compare the economic well-being and governance quality of member and nonmember countries while also comparing within the counties before and after membership was secured. Using the panel data, several ordinary least square (OLS) regression models will be estimated with dependent variables measuring economic wellbeing and several indicators of quality of governance. Due to missing data, the results for GDP per capita contain 186 countries and results for the governance variables contain 196 countries. All variables used are described along with their sources in ‘Appendix A: Variables'. The statistical model that will be estimated allows the examination of the interaction between the two main independent variables, EITI Membership and the amount of natural resources within the country (RES). The equation appears as follows: Yit ¼ β0 þ β1 RESit þ β 2 EITI_MEMit þ β 3 RESit  EITI_MEMit þ β4 Zit þ ε where: Yit ¼ The dependent variable for country i at time t. RESit ¼The independent variable for resource abundance for country i at time t. EITI_MEMit ¼The independent variable for EITI

5

membership (1 ¼ member, 0 ¼ non-member) for country i at time t. Zit ¼ A vector of control variables for country i at time t. ε ¼Error— random, normally distributed, and independent.

Independent variables Membership in the EITI is the primary independent variable of interest in this study. A data set for EITI membership was constructed solely for the purposes of this paper from qualitative information given on the EITI webpage. A dummy variable was used for all countries from 1995 to 2009, where for each year each country was assigned a “0” if they were not an EITI “member” or a “1” if they were a “member”. The variable for EITI membership (EITI_MEM) was defined as the point where a country expressed intention to become a member state. When no date was given for country intention, the date of candidacy approval was used for this variable. Some countries specifically announced their intention to work towards candidacy while others did not. Announcing candidates were recorded by the EITI. Since candidacy requires announcing a clear statement of the government's commitment, developing a work plan setting objectives for what a country wants to achieve with their EITI and how it intends to reach Compliant Status, and establishing a multi-stakeholder group (MSG) together with companies and civil society(Extractive Industries Transparency Initiative, 2013b), the EITI_MEM variable indicates either accomplishment of all three requirements or recorded intention by the country in question to pursue all three requirements. The advantage of identifying this specification is having the longer time spans available to look at “membership” effects and the assumption that even before being accepted as a candidate country, demonstrated intention implies a willingness to change transparency policies and work towards EITI membership requirements.2 Although EITI_MEM takes the preparation process or the “intention” to join into account to an extent, it cannot fully capture all polices or plans toward increasing transparency and accountability. Some caution, therefore, should be taken in interpreting the results which are based heavily on the exact timing of the membership variable. In this case, comparisons between countries may be more revealing than comparisons within a country over time. The second independent variable is a proxy for natural resources in a country. This is of key interest because the resource curse is based on the observation that an amount of natural resources over a certain threshold can have a negative effect on nations under certain circumstances. Although the EITI mainly deals with oil and mineral resources, countries such as Liberia have chosen to include the timber industry within their EITI compliance as well. The first measure, labeled RES, is, therefore, a measurement of the total primary exports divided by total merchandise exports. The data were taken from the UNCTAD (United Nations Conference on Trade and Development, 2002).

Dependent variables The dependent variables have been chosen to reflect the effect of membership in EITI on nations' economic development and quality of governance. The first regression model will use GDP per 2 Additionally EITI_MEM_2 and EITI_MEM_3 were constructed. Where EITI_MEM_2 equals “1” only once a country has been declared a candidate by the EITI. When no exact date was given for compliant countries, candidacy was assumed to be given 2 years before compliance, as this is stated in the rules of the EITI as the standard amount of time allowed to move from candidacy to compliance. EITI_MEM_3 is the strictest measurement of EITI “membership”. “1” only corresponds to the year where EITI countries were declared compliant. However, this measure is problematic because of the relatively small number of compliant states and the fact that none were given compliance status before in 2010 and many were given compliance status only in March 2011. A limitation of EITI_MEM_2 and 3 is that they treat membership as an event, when it is probably truly a process of preparation and implementation over a number of years surrounding the event of membership.

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

Table 1 Dependent variables. Dependent variables Economic indicators

Gross domestic product per capita

Governance indicators

Based on The Worldwide Governance Indicators (WGI) project at the World Bank (1996–2009) (Kaufmann et al., 2010) Voice and accountability Political stability no violence Government effectiveness measure Regulation quality Rule of law Control of corruption

capita as the dependent variable. These data were obtained from the World Development Indicators compiled by the World Bank. The governance measures are based on indicators from Kaufmann et al. (2010), found in The Worldwide Governance Indicators (WGI) project at the World Bank. Data were gathered for the WGI project from 31 different sources. These include companies which had conducted surveys, public sector data providers, nongovernmental organizations, and commercial business information providers. The WGI project then compiled and scaled the scores from around 2.5 to 2.5 on a county/year basis for each of the six indicators. Kaufmann et al. (2010) define governance as “the traditions and institutions by which authority in a country is exercised”. This includes: (a) the process by which governments are selected, monitored and replaced; (b) the capacity of the government to effectively formulate and implement sound policies; and (c) the respect of citizens and the state for the institutions that govern economic and social interactions among them (4). This project measures six aspects of governance, namely voice and accountability (democracy measure), political stability/no violence (stability and absence of conflict measure), government effectiveness (quality of public service measure), regulation quality (quality of policy measure (esp. private sector development)), rule of law (ability to abide by and predictability of law/rules measure (esp. property rights)), and control of corruption (perception of corruption measure). Under the definition, voice and accountability and political stability/no violence belong to category (a), government effectiveness and regulatory quality belong to category (b), and rule of law and control of corruption belong to category (c). All six serve as dependent variables in separate models (Table 1).

Control variables Additionally, several control variables are included in the models. The data sources and definitions of the control variables are listed in Appendix A. The control variables are factors that are known to correlate with the dependent and independent variables. They are included in the regression model in order to reduce the chances of omitted variable bias. For the dependent variable, GDP per capita, inflation, investment, government consumption, democracy levels, population and openness in terms of trade were controlled for. For the governance indicators population, GDP per capita growth and government consumption were used in most cases. Additionally, democracy was controlled for except when using the indicators from category (a) (because these are already measures of democracy to an extent). Conflicts were control for when using the dependent variable rule of law and voice and accountability (since these indicators would be highly affected by conflict). Lastly, education was controlled for when looking at

Based on constant 2000 US dollars. Source: World Development Indicators at the World Bank.

Measurement of level and quality of democracy within a country Measurement of the stability of a government as well as the absence of conflict Measurement of the quality of public service Measurement of the quality of policies made within a country Measurement of the ability to abide by or the predictability of law within a country, this especially concerns property rights Measurement of perception of the control of corruption within a country

political stability/no violence. Unfortunately, it is not possible to control for unobserved factors that might have affected selection into EITI membership, but by adding measured control variables to the model, some of the known sources of bias are taken into account.

Hypotheses While there is still much debate on the issue of the resource curse, it can be generally suggested that in countries with weak institutions, the more natural resources a country has, the worse off it will be economically. Additionally, a link between natural resource abundance and reduced quality of democracy, increase of corruption, and higher chances of conflict (depending on the resource in question) has been suggested. This paper, therefore, makes the assumption that natural resources have a negative effect on both economic development and governance in countries with weak institutions. The EITI aims to increase transparency and accountability in order to improve institutional quality. Since studies have shown that in countries with strong institutions, natural resource abundance actually increases growth rather than decreases it, it can be assumed that attempts to increase the quality of institutions through transparency and accountability would improve economic development (measured here in GDP per capita). Furthermore, through strengthening institutions, governance in EITI member states should also improve. In theory, by requiring financial disclosure by governments of profits earned in the natural resource sector, less money can be misappropriated by government officials, which should result in decreased corruption and an increased rule of law, improved quality of policies and more effective decision-making. Government officials will also have less reason to be undemocratic given that the office does not yield the same personal wealth that it did before (less possible corruption), and citizens will have fewer grievances based on natural resource sector earnings staying in the hands of corrupt officials, also resulting in direct or indirect improvements of the process by which governments are selected, monitored and replaced. Two hypotheses will be examined, empirically, in this article, using the methods discussed above. They are as follows: Hypothesis 1:. Depending on the size of the resource sector within a country, EITI membership will improve economic development (measured in GDP per capita). Hypothesis 2:. Depending on the size of the resource sector within a country, EITI membership will improve governance within a country (measured by six different indicators: voice and

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

accountability, political stability/no violence, government effectiveness, regulation quality, rule of law and control of corruption). If the positive effects from EITI membership are great enough, as measured by the coefficient of the interaction term, they should be able to counteract (or counterbalance) or at least mitigate some of the negative effects of resource abundance on the dependent variables, as measured by the coefficient of the RES term. Model assumptions, specifications, and missing data To address the problem of serial autocorrelation, the regression analysis utilized robust standard errors. The residuals were examined in order to check for normality without extreme outliers and equal variances. When necessary, the log of the variable was taken to help address departures from normality. Furthermore, unlike controlled experiments, panel studies are vulnerable to extraneous influences that are unmeasured. Fixed effect models are a frequently used method to compensate for this bias. In this analysis, all the models used time-fixed effects by year to correct for omitted variable bias due to the occurrence of unmeasured changes across time that are constant among countries. Country-fixed effects were utilized only in the final model for each dependent variable and should be interpreted carefully because of the limitations of the EITI membership variable. For some of the countries, EITI membership in reality may have been non-time varying because it takes several years to get ready to join and it takes several years to implement, and even “intention” to join could not capture all of this. The panel of eleven to fifteen years is relatively short to capture in a country-fixed effects model, a process that occurs over several years and would leave out the important between member and non-member comparisons that the study also aims to analyze. Additionally, because for some countries, membership is recent and policy may not have been fully implemented at the time of the study, the effects might not be seen over time within the country itself. Due to the fact that EITI activities may have been non-time varying for many countries during this study, a country fixed effects model is not used in that it would be unable to detect the influence of membership. For the purposes of this study, emphasis was placed on providing all EITI member states with maximum data possible to minimize the cases that had to be eliminated from the study due to missing data. The three countries with the largest gaps in data were Afghanistan, Iraq, and Timor-Leste. Indonesia also had minor gaps. When occasional single year, single indicator gaps were found, the missing data were filled in by taking the value for the closest year for the same country and same indicators. Additionally, some missing data were found in an alternative data set, mainly the IMF World Economic Outlook Database (International Monetary Fund, 2010). This again was done only for countries in the EITI member state group as it was very important to retain this relatively small subset of countries in the analyses. Results Seven separate regression analyses were carried out to examine the effect of EITI membership on economic development as well as aspects of governance quality. A model building approach was adopted, beginning with a simple regression that included only one of the seven dependent variables and the measurement for resource abundance, here RES. The third to last model includes all appropriate controls along with the interaction term: RESnEITI_MEM. The second to last model uses country-fixed effects in addition to time—fixed effects. However, due to the short time period, and the fact that country-fixed effects exclude the between country differences that are of concern to the hypotheses, these results are not extensively

7

commented on. The final model controls for Norway (because it is an outlier in terms of wealth and governance), the results of which are discussed in “Alternative measurements” section. Because of the use of the interaction term in Models 3 through the last model for each dependent variable, the main effect coefficients cannot be interpreted across the board, but must take the interaction term coefficients into account. Therefore, the significance of the effect of resource abundance and EITI membership alone on the dependent variable should be considered in models one and two but not in the interaction models (Allison, 1999). Gross domestic product per capita GDP per capita was the indicator used to measure economic well-being within a country. The hypothesis that EITI membership reduces the negative effect of natural resources on GDP per capita was supported. Table 2 presents the results for successively more complex models. Model 1 suggests that RES does have a significant negative effect on GDP per capita, as would be expected by the resource curse. Model 2, shows also that EITI membership has a negative effect on GDP per capita, suggesting that countries in the process of becoming EITI compliant tend to have lower GDP. The fact that EITI membership has a negative coefficient in all of the models suggests the possibility of negative selection, the complicated nature of the EITI membership process or short-run problems with the effectiveness of the EITI. These are also possible explanations for the main negative main effect of membership observed in all of the governance indicators. However, a more thorough investigation as to the cause of the negative main effect of EITI is still needed. Models 3 through 12 include the interaction term between EITI membership and RES. The significant and positive interaction coefficient in Models 3 through 10 suggests that EITI membership reduces the negative influence of RES on GDP per capita, consistent with diminishing the resource curse (in model 10 the coefficient of the interaction term suggests that joining EITI adds 3.48 to the coefficient for RES on GDP per capita).3 Governance quality There are six indicators of governance quality that are examined in this section and the statistical modeling follows the same strategy as reported above for GDP per capita. The results for the dependent variable, government effectiveness, are reported Table 3. Model 1 where RES is entered alone shows that resource abundance has a negative and significant effect on government effectiveness. Model 2 shows again the negative main effect of EITI membership. Models 3 through 8 display that EITI membership RES interaction has a significant and positive influence on government effectiveness. The strength of the coefficient suggests that the EITI's influence is strong enough in this case to “defeat” the resource curse with EITI membership, increasing the benefits of RES for government effectiveness. Regulation quality is concerned especially with measuring the private sectors ability to develop. In Table 4, Model 1, once again, displays that increased resource abundance has a negative effect on regulation quality (Model 2 again displays the negative main effect of the EITI membership). This is in keeping with the findings from the literature: that regulatory quality may be vulnerable in resource abundant states where the government is not sufficiently accountable to its citizens. The effect of EITI membership on regulation quality remained positive and significant under all 3 Given the coefficient of  1.47 for RES, EITI membership, in this case, changes the influence of RES on GDP per capita to a positive value. In sum being an EITI member should improve the influence of RES on GDP per capita to 2.01 (  1.47 þ3.48).

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

8

EITI membership effect on GDP per capita models

RES

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11) Cotunry fixed effects

(12) Control for Norway

 2.17 (  6.39)nn

 2.07 (  6.05)nn  .88 (  3.22)nn

 2.13 (  6.17)nn  3.98 (  5.36)nn 3.77 (3.86)nn

 1.82 (  5.65)nn  3.53 (  6.02)nn 3.21 (4.10)nn  .45 (  7.53)nn

 2.14 (  6.03)nn  3.67 (  3.88)nn 3.61 (2.84)nn

 2.04 (  6.11)nn  3.85 (  3.49)nn 4.34 (3.09)nn

 1.86 (  4.89)nn  4.08 (  5.44)nn 4.11 (3.97)nn

 2.32 (  7.04)nn  4.24 (  5.84)nn 4.24 (4.62)nn

 1.98 (  5.63)nn  3.90 (  4.39)nn 3.73 (3.28)nn

 1.47 (  4.24)nn  3.08 (  3.36)nn 3.48 (2.92)nn  0.39 (v6.20)nn 0.33 (1.43)

.001 (  0.12)  0.09 (  0.47) 0.13 (0.51)  0.01 (v0.94) 0.04 (0.66)

 1.56 (  4.57)nn  3.17 (  3.46)nn 3.58 (3.00)nn  0.37 (6.09)nn 0.33 (1.42)

0.94 (3.35)nn 0.06 (4.05)nn  0.04 (  0.55)  0.0 (  0.00)

 0.15 (  1.17) 0.002 (0.85)  0.69 (  3.12)nn  0.02 (  0.40)

EITI_MEM RESnEITI_MEM Log(INFLATE)

0.55 (2.42)n

Log(INVEST)

1.16 (3.94)nn

LOG(GOVT_CONSUMP)

0.06 (3.09)nn

POL2

 0.18 (  3.95)nn

LOG(POP)

0.54 (2.53)n

LOG(OPEN)

2.19 (11.53)nn

Norway Country included in Model 3: 186. T-statistic in parenthesis ( þ o 0.10, n o 0.05,

0.91 (3.23)nn 0.06 (3.85)nn  0.42 (  0.52) 0.001 (0.00)

nn

o 0.01 significance).

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

Table 2 Results log of GDP per capita, using robust standard errors and time fixed effects.

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

9

Table 3 Results government effectiveness, using robust standard errors and time fixed effects. EITI membership effect on government effectiveness models

RES

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9) Country fixed effects

(10) Control for Norway

 1.57 (  8.15)nn

 1.49 (  7.65)nn  0.51 (  3.99)nn

 1.52 (  7.70)nn  1.70 (  5.49)nn 1.46 (3.71)nn

 1.43 (  7.76)nn  1.36 (  2.71)nn 1.37 (2.24)n 0.72 (5.08)nn

 1.23 (  5.35)nn  1.74 (  5.14)nn 1.62 (3.67)nn

 1.58 (  8.22)nn  1.75 (  5.47)nn 1.60 (4.01)nn

 1.50 (  7.81)nn  1.53 (  4.41)nn 1.42 (3.15)nn

 1.12 (  5.26)nn  1.37 (  2.38)n 1.50 (2.08)n 0.77 (5.57)nn 0.04 (4.87)nn  0.01 (  0.22)  0.062 (1.86)

 0.07 (  0.50) 0.18 (1.06)

 1.17 (5.63)nn  2.37 (  2.41)n 1.52 (2.14)n

EITI_MEM RESnEITI_MEM LOG(GOVT_CONSUMP) POL2

0.05 (4.76)nn  0.80 (  3.12)nn

LOG(POP)

 0.13 (  3.28)nn

LOG(GDPPC_GROWTH)

 0.26 (  1.28) 0.02 (0.41) 0.004 (0.75)

0.04 (4.67)nn

 0.50 (  1.54) 0.01 (1.38)

 0.01 (  0.16)  0.06 (  1.73) þ 1.58 (13.09)nn

Norway

Country included in Model 3: 196. T-statistic in parenthesis ( þ o 0.10, n o 0.05,

nn

0.74 (5.42)nn

o 0.01 significance).

Table 4 Results regulation quality, using robust standard errors and time fixed effects. EITI membership effect on regulation quality models

RES

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8) Country Fixed Effects

(9) Control for Norway

 1.55 (  8.47)nn

 1.51 (  8.02)nn  0.28 (  2.11) n

 1.55 (  8.12)nn  1.61 (  3.79)nn 1.63 (3.30)nn

 1.05 (  5.14)nn  1.64 (  3.54)nn 1.73 (3.04)nn 0.07 (6.10)nn

 1.61 (8.63)nn  1.66 (  3.82)nn 1.77 (3.45)nn

 1.48 (  8.16)nn  1.41 (  2.99)nn 1.50 (2.64)nn

 1.10 (  5.63)nn  1.65 (  3.24)nn 1.79 (2.89)nn 0.06 (6.45)nn  0.06 (  1.84) þ  0.11 (  3.23)nn

0.10 (0.64) þ

 1.13 (  5.81)nn  1.65 (  3.24)nn 1.81 (2.91)nn

þ

0.06 (6.27)nn

EITI_MEM RESnEITI_MEM POL2

 0.08 (  3.20)nn

LOG(POP)

 0.16 (  3.76)nn

LOG(GDPPC_GROWTH) Norway Country included in Model 3: 196. T-statistic in parenthesis ( þ o 0.10, n o 0.05,

nn

0.39 (1.92)  0.40 (  1.65) 0.01 (1.89)  0.62 (  1.42) 0.01 (0.45)

 0.06 (  1.79)  0.11 (  3.16)nn 1.03 (10.49)nn

o 0.01 significance).

controls (Models 3–7). In all cases, the coefficient was also strong enough to counteract the negative effect of resource abundance. Rule of law measures the reliability of law enforcement within a country. This especially concerns the quality of property rights. In Table 5, Model 1 displays a resource curse effect for rule of law that is negative and significant (Model 2 again displays the negative main effect of EITI membership). Models 3 through 8 display that EITI membership has a positive and significant interaction effect with RES on rule of law within a country. The coefficient is strong enough in all models, except Models 3 and 6, to “defeat” the resource curse for rule of law. Voice and accountability (VA) is a measurement of level and quality of democracy within a country. The significance of the coefficient for the RES–EITI interaction varies depending on the models. An examination of Table 6 shows that VA is significantly lowered as resource abundance increases. Model 2 displays however, that EITI membership does not significantly affect VA. The interaction results are significant at the 0.10 alpha level only when controlling for conflicts or population but the interaction term

coefficients are less positive than the RES coefficient is negative. Being an EITI member, therefore, does not improve the influence of RES on voice and accountability enough for counteracting the resource curse. Political stability/no violence is a measurement of the stability of a government as well as the absence of conflict. The increased likelihood for these types of activity in resource abundant states is also documented in the literature and confirmed in the first two model of Table 7 with the significance of RES. Table 7 also shows that EITI membership appears to have no significant relationship political stability/no violence. Control of corruption is a measurement of perception of the control of corruption within a country. In Table 8, results show that resource abundance does have a negative effect on the control of corruption in concert with the literature reviews (Model 1). Furthermore, as seen with the other indicators, EITI membership alone has a negative effect as well (Model 2). The significance of the relationship and the strength of the coefficient change and are weak for the interaction effect of EITI membership and RES on

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

10

Table 5 Results rule of law, using robust standard errors and time fixed effects. EITI membership effect on rule of law models

RES

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9) Country fixed effects

(10) Control for Norway

 1.50 (  7.64)nn

 1.42 (  7.14)nn  0.56 (  4.25)nn

 1.46 (  7.19)nn  1.74 (  6.21)nn 1.44 (3.84)nn

 1.35 (  7.26)nn  1.38 (  2.86)nn 1.47 (2.47)n

 1.20 (  5.27)nn  1.69 (  5.91)nn 1.59 (3.95)nn

 1.41 (  7.10)nn  1.50 (  5.11)nn 1.34 (3.33)nn

 1.47 (  6.78)nn  1.74 (  6.24)nn 1.57 (4.43)nn

 0.15 (  1.19) 0.13 (0.52)

 1.06 (  5.44)nn  1.41 (  2.87)nn 1.58 (2.58)n

 0.63 (  5.21)nn

 1.00 (  5.02)nn  1.41 (  2.82)nn 1.56 (2.50)n 0.77 (5.40)nn 0.04 (4.62)nn  0.07 (  1.79) þ  0.33 (  3.35)nn

EITI_MEM RESnEITI_MEM LOG(GOVT_CONSUMP)

0.82 (5.91)nn

POL2

0.05 (4.49)nn  0.15 (  3.52)nn

LOG(GDPPC_GROWTH) CONFLICTS

 0.16 (  0.55) 0.05 (1.01)

0.75 (5.28)nn

0.01 (2.41)n

0.04 (4.40)nn

 0.005 (  0.70)  0.03 (  0.70)

 0..06 (  1.68) þ  0.32 (  3.24)nn 1.56 (12.34)nn

Norway

Country included in Model 3: 196. T-statistic in parenthesis ( þ o 0.10, n o0.05,

nn

o 0.01 significance).

Table 6 Results voice and accountability, using robust standard errors and time fixed effects. EITI membership effect on voice and accountability models

RES EITI_MEM RESnEITI_MEM

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9) Country fixed effects

(10) Control for Norway

 1.38 (  7.29)nn

 1.35 (  6.97)nn  0.26 (  1.90) þ

 1.35 (  6.90)nn  0.55 (  2.01)n 0.36 (0.96)

 1.33 (  6.94)nn  0.28 (  0.66) 0.32 (066)

 1.49 (  7.25)nn  0.70 (  2.66)nn 0.69 (1.90) þ

 1.49 (  8.32)nn  0.67 (  2.45)n 0.66 (1.71) þ

 1.31 (  6.64)nn  0.32 (  1.13) 0.30 (0.75)

 1.43 (  7.13)nn  0.30 (  0.77) 0.53 (0.98)

 0.82 (  0.61) 0.41 (1.64)

 1.46 (  7.40)nn  0.31 (  0.81) 0.55 (1.04)

 0.16 (  3.60)nn

0.56 (3.80)nn  0.26 (  1.99)n  0.07 (  1.43)  0.13 (  2.73)nn

LOG(GOVT_CONSUMP)

0.61 (4.08)nn  0.59 (  5.08)nn

CONFLICTS

 0.16 (  6.32)nn

LOG(POP) LOG(GDPPC_GROWTH) Norway

Country included in Model 3: 196. T-statistic in parenthesis þ o 0.10, n o 0.05,

nn

 0.39 (  1.31) 0.08 (1.95)

þ

 0.12 (  2.15)n  0.35 (  1.44) 0.02 (2.50)n

0.54 (3.65)nn  0.25 (  1.89) þ  0.07 (  1.44)  0.12 (  2.63)nn 1.61 (14.84)nn

o 0.01 significance).

control of corruption. The results are not highly significant in the final model and the coefficients are never strong enough to counteract the resource curse effect.

Alternative measurements Alternative measures were constructed from the World Bank's World Development Indicators. This measure (FOM for fuels, ores, and metals) uses only the fuel, ore and metal exports as a percentage of total exports. However, while this measure is more specific to the resources targeted by the EITI, the data were more incomplete, yielding a smaller sample size. Additionally, this variable had a number of cases that were outliers and a high coefficient of variation of 1.28. For RES the coefficient of variation was moderate (.55).

The effect of the interaction term remained significant only for GDP per capita when using this indicator. The results for regulation quality, rule of law, and government effectiveness were sensitive to this measure. This sensitivity test does not lessen the importance of the results reported using the RES indicator. However, while keeping in mind the disadvantages of the FOM measurement, mentioned, the use of FOM has changed the results to an extent. This goes in keeping with the cautions mentioned in the literature review: that using different measures of resource abundance can change, not only the results in this case, but can also start a debate as to whether the resource curse exists at all. Additionally, models were run using Total_Rents as the resource abundance indicator. These data were taken from the World Bank Development Indicators and are the measure of oil, natural gas,

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

11

Table 7 Results political stability/no violence, using robust standard errors and time fixed effects. EITI membership effect on political stability/no violence models

RES

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8) Cotunry fixed effects

(9) Control for Norway

 1.02 (  4.90)nn

 0.95 (  4.52)nn  0.56 (  3.16)nn

 0.96 (  4.55)nn  1.15 (  2.43)n 0.71 (1.09)

 0.42 (  2.15)n 0.22 (0.32)

 1.16 (  6.43)nn  1.31 (  2.93)nn 1.16 (2.06)n

 0.88 (  4.22)nn  0.85 (  1.72) 0.52 (0.75)

 0.74 (  4.38)nn  0.23 (  0.38) 0.25 (0.34)

 0.33 (  2.00)n 0.23 (0.50)

 0.77 (  4.56)nn  0.27 (  0.44) 0.29 (0.41)

 0.07 (  1.91)

0.01 (7.07)nn  0.20 (  10.15)nn 0.09 (  2.81)nn

EITI_MEM RESnEITI_MEM SECONDEDU2

 0.52 (  0.63) 0.02 (8.08)nn

 0.24 (  11.27)nn

LOG(POP) LOG(GDPPC_GROWTH)

 0.34 (  0.66) 0.004 (1.80)  0.28 (  0.76) 0.01 (1.01)

Norway Country included in Model 3: 196. T-statistics in parenthesis (þ o 0.10, n o 0.05,

nn

0.01 (6.77)nn

þ

 0.20 (  10.19)nn  0.09 (2.73)nn 0.80 (8.07)nn

o 0.01 significance).

Table 8 Results control of corruption, using robust standard errors and time fixed effect. EITI membership effect on control of corruption models

RES EITI_MEM RESnEITI_MEM

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9) Country fixed effects

(10) Control for Norway

 1.38 (  6.78)nn

 1.31 (6.33)nn  0.52 (  4.56)nn

 1.33 (  6.31)nn  1.30 (  4.68)nn 0.95 (2.70)nn

 1.05 (  4.23)nn  1.25 (  4.22)nn 1.04 (2.73)nn 0.05 (4.08)nn

 1.42 (  7.09)nn  1.38 (  4.74)nn 1.17 (3.23)nn

 1.22 (  6.20)nn  0.92 (  1.72) 0.91 (1.50)

 1.32 (  6.26)nn  1.06 (  3.90)nn 0.87 (2.40)n

 0.97 (  4.25)nn  0.92 (  1.67) þ 1.03 (1.62)

 0.13 (  0.94) 0.23 (1.13)

 1.03 (  4.57)nn  0.92 (  1.70) þ 1.06 (1.68)

POL2

 0.12 (  4.81)nn

LOG(POP) LOG(GOVT_CONSUMP)

0.91 (6.68)nn  0.16 (  3.80)nn

LOG(GDPPC_GROWTH) Norway

Country included in Model 3: 196. T-statistic in parenthesis ( þ o 0.10, n o 0.05,

nn

0.04 (4.35)nn  0.05 (  1.53) 0.88 (6.24)nn  0.10 (  2.63)nn

þ

 0.33 (  1.41)  0.01 (1.43)

0.04 (4.13)nn

 0.12 (  0.31) 0.09 (1.26)

 0.05 (  1.48) 0.86 (6.10)nn

0.02 (2.36)n

 0.09 (  2.53)n 1.63 (12.12)nn

o 0.01 significance).

and mineral rents as a percentage of GDP. The use of rents instead of exports has the advantage of not being distorted by the fact that more developed countries tend to use more of their resource domestically (Alexeev and Conrad, 2005). However, this proxy also had a high coefficient of variation (2.34), some missing data, and outliers. Therefore, for the purposes of this study, a broader index of natural resource exports is the most practical and available measurement. In this case, government effectiveness and regulation quality were the only indicators that remained significant (government effectiveness only at a significance level of 0.10). GDP per capita and rule of law were no longer significantly affected. Additionally, the variable EITI_MEM_2 was also used to determine whether results remained unchanged under a more narrow definition of EITI membership. Again, using all appropriate variables (i.e. the final model for each dependent variable), the significance of the interaction term remained unchanged for all seven dependent variables. Finally, because Norway, with very high-quality governance and economic development, is an outlier in terms of EITI

membership, the final model for each dependent variable controlled for Norway. There was no effect on significance for GDP per capita, rule of law or political stability. Regulation quality, voice and accountability, government effectiveness and the control of corruption models had slight changes in the significance of the control variables, but not the variables of interest. Only control of corruption experienced a slight change in significance of the EITI interaction variable, but is still not highly significant (controlling for Norway, the interaction effect was significant at a 0.10 alpha level).

Analysis While the debate on the existence of the resource curse remains, transparency and accountability for resource-rich countries continues to be an important and current topic. In terms of oil and minerals, the EITI has become the major voice for establishing standards for transparency and accountability in resource rich countries.

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

12

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

A statistical analysis of approximately 200 countries shows that the statistical interaction between resource abundance and EITI membership has a significant and positive effect in terms of moderating the effect of natural resources abundance on GDP per capita, government effectiveness, regulation quality and rule of law in EITI member countries in comparison with non-member countries and EITI member countries before they became members. The strength of the effect is strong enough in many instances to bring the negative effects of resource abundance (as measured by RES) close to 0 or even move the coefficient to become positive. The buffering effect of EITI membership on the negative influence of RES on GDP per capita is the strongest out of all regression analyses conducted for this study. The effect of resource abundance on economic development is the main focus of research on the resource curse and therefore a noteworthy finding in terms of the success of the EITI's goals. It is important to note that this study was not able to rule out unmeasured factors that may influence decisions to join EITI, even though the EITI variable was expanded to include “intention” and time fixed effects were used in all models. This is suggested by a negative coefficient on the EITI membership's main effect for which negative selection, the complicated nature of the EITI membership process and/or short-run problems with the effectiveness of the EITI, could be possible explanations that need further investigation. Furthermore, sensitivity tests demonstrated that alternative measures for resource abundance yielded some variation in findings. While these alternative measures have their own flaws, their effect on the results should be noted and also point to the fact that there is a need for further investigation within this area of research. On a broader scale, this paper divides the negative effects of resource abundance into economic wellbeing and quality of governance effects. The governance effects can be further divided into policy making quality, the quality of citizen's respect for institutions and democratic quality. According to the results, when resource abundance is substantial enough (keeping in mind the negative EITI main effect), EITI membership has had a positive effect on (1) economic development and (2) what Kaufmann et al. (2010) describe as “the capacity of the government to effectively formulate and implement sound policies (4)”. This means that, in concert with the goals of EITI membership has perhaps prompted the creation of more useful policies and policy making. Additionally, the significant increase in GDP per capita since membership could be a response to more development-oriented policies. These findings support the EITI's goal of having resources benefit all in a society and increasing accountability involving resources. But it should still be used with caution for countries considering joining the EITI, as the circumstances depend on the magnitude of resource abundance. Because of the weak direction and changing significance of the results for control of corruption, and the significant effects of EITI membership on rule of law, membership has an undetermined effect on what Kaufmann et al. (2010) describe as (3) “the respect of citizens and the state for the institutions that govern economic and social interactions among them (4).” This could mean that while EITI membership seems to have improved policy making and the quality of policy, it has not uniformly improved citizen's trust in institutions. Perhaps while EITI membership has fostered transparency in these areas, the means of informing citizens of the improvements has not been adequate; or, it could be that the EITI has only fought corruption in limited areas but not in all areas. In addition, since this panel study had only a few data points occurring post-EITI membership, more time may need to pass before citizens are affected by the improved transparency. Again, more research is needed in this area.

Membership seems to have had no significant effect on what Kaufmann et al. (2010) describe as (4) “the process by which governments are selected, monitored and replaced (4)”. Both the measure for democratic quality (voice and accountability) and stability (political stability/no violence) had wavering significance and weak direction. In the literature review, it was argued that natural resource abundance often undermines democracy by allowing more incentives for patronage politics and corruption (Collier, 2007). Since the effect on corruption by EITI membership is also weak, it could be inferred that the inability to deter corruption permits these opportunities to remain. This could also point to the distinction between transparency and accountability. While EITI membership may have improved transparency, the improvements in accountability may not be as determinative. This could further explain the insignificance of the effect of EITI membership on political stability and a lack of violence. It may be impossible within the limits of this study to factor out all other causes for conflict or government instability but it might be that the EITI has not been successful in eliminating the grievances or greed associated with resource abundance. Additional research in this area should include a focus on trying to distinguish different measures of democratic quality to parse out which factors are more or less affected by membership or, alternatively, to focus on distinguishing between different forms of conflict or government change, to see which types of violence or instability are more associated with EITI membership. Furthermore, it is important to note that the significant changes associated with EITI membership are not simply isolated effects, but rather occur by lowering the negative effects of growing resource abundance on the economic and governance indicators. Thus, the EITI itself is not a panacea. Results demonstrate a negative main effect of EITI membership, suggesting that countries without sufficient resource abundance might not experience benefits from a decision to join. A different explanation is that EITI membership initially has a negative influence resulting from the added bureaucratic burden yet in the long-run the influence becomes positive. Additionally, it is not clear how events and conditions in the years prior to joining the EITI may have influenced these findings. Therefore, these analyses should be repeated five to ten years hence when EITI membership has had a longer time to fully affect governance and more long-term analysis of GDP per capita can be undertaken. Qualitative studies are also needed to determine the factors that influence the choice to join the EITI, the processes that lead up to such a decision and have additional membership effects that are not observable in the empirical study. With those factors in mind the herein results merely provide a quantitative look at the initial successes of the EITI and suggest areas for further investigation. Furthermore, since the time period that the analysis covers, up to 2009, several countries have gained their candidacy, while and several others have been suspended.4 Rerunning the analysis in light of these events (once the dependent variable data has also been made more current) could shed more light on the effects of the EITI.

4 Chad, Honduras, Tajikistan, The Philippines, Trinidad and Tobago, Togo and Guatemala have become EITI candidates since the time period of the original analysis ended. Equatorial Guinea lost its candidacy in 2010, Gabon lost it candidacy in 2013. Madagascar was suspended in 2011 and Yemen, Sierra Leon, the Democratic Republic of the Congo, The Republic of the Congo, and the Central Africa Republic all have been suspended as of 2013. Additionally, San Tome and Principe was a candidate from 2008 to 2010 and became a candidate again in 2012 (Extractive Industries Transparency Initiative, 2013).

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

13

Table A1 Variables. Code

Description

Independent variables EITI_MEM EITI_MEM_2 EITI_MEM_3 RES FOM Total_Rent Dependent variables GDPPC Governance indicators V_A PolStab Govt_Effect REGQUAL ROL ConCorupt Control variables INVEST INFLATE OPEN Govt_Consump POP SECONDEDU POL2

CONFLICTS GDPPC_Growth Norway

Dummy variable for intention to enter EITI (1 ¼ intention, 0¼ no intention), measured annually. Based on EITI Website Dummy variable for candidate to enter EITI (1 ¼candidate, 0¼ non-candidate), measured annually. Based on EITI Website Dummy variable for compliant EITI member (1 ¼ member, 0¼ non-member), measured annually. Based on EITI Website Oil/mineral proxy, measured as primary exports/total exports. Based on data from United Nations Conference on Trade and Development (2002) Oil/mineral proxy. Measures as fuel, ore, and metal exports(% of merchandise exports) Oil, natural gas, and mineral rents (% of GDP) Gross domestic product per capita (constant 2000 US$) Based on The Worldwide Governance Indicators (WGI) project at the World Bank (1996–2009) (Kaufmann et al., 2010) Voice and accountability (democracy measure) Political stability no violence (stability and absence of conflict) Government effectiveness measure (quality of public service) Regulation quality (quality of policy) Rule of Law (ability to abide by and predictability of law/rules (esp. property rights)) Control of corruption (perception of corruption)

Gross capital formation (% of GDP) Inflation, consumer prices (annual %) Measure of openness ¼ Imports of goods and services (% of GDP) þ Exports of goods and services (% of GDP) General government final consumption expenditure: includes all government current expenditures for purchases of goods and services Total population, based on defacto definition of population School enrollment, secondary (% net) Combined Polity Score: Computed by subtracting AUTOC from DEMOC; normal range polity scores are imputed for coded “  77” and “  88” special polity conditions, polities coded “  66” on the POLITY variable are left blank. Based on PolityIV Project (Marshall et al., 2010) Incidence of intrastate conflict. Coded 1 in all country-years with at least one active conflict. Based on UCDP/PRIO Armed Conflict Dataset v.4–2010 (Department of Peace and Conflict Research, Uppsala Universitet, 2010). Gross Domestic Product per capita growth (annual %) Dummy variable for Norway

n

Unless specified, data is based on World Bank Development Indicators (The International Bank for Reconstruction and Development and the World Bank, 2011).

Table B1 Lists of EITI countries (as of 2009). Afghanistan Albania Azerbaijan Burkina Faso Cameroon Central African Republic Congo, Dem. Rep. Congo, Rep. Cote d'Ivoire Equatorial Guinea Gabon Ghana Guinea Indonesia Iraq Kazakhstan

Kyrgyz Republic Liberia Madagascar Mali Mauritania Mongolia Mozambique Niger Nigeria Norway Peru San Tome and Principe Sierra Leone Tanzania Timor-Leste Yemen, Rep. Zambia

Conclusion While EITI membership appears to have helped countries improve in terms of one of its major goals, namely, allowing resources to benefit all and improving transparency (when measuring benefits in GDP per capita and policy making quality), the goals of stabile democratic systems and lower corruption remain elusive. The EITI's success, tentatively shown to some extent in this study, is, at best, inconclusive as the positive effects are apparently conditional on having sufficient resource abundance, coupled with an incomplete picture of the context and conditions surrounding the process of membership and implementation. However, these findings should not deter the greater

mission of making the resource sector more transparent and holding governments more accountable. The literature represents the reality that resource-rich countries struggle more to keep checks and balances in place and to keep large resource rents transparent and beneficial to the population. An international initiative to combat this may not be the only solution, but it is clear that resource-rich countries can benefit from assistance to combat these problems effectively. Time may demonstrate that the EITI can help to minimize the negative effects of the resource curse through improved transparency and accountability to all who wish to join.

Acknowledgements I would like to thank Dr. Mohammad Reza Farzanegan and Dr. Matthias Basedau for their help and guidance in structuring and carrying out my research. Additionally, I would like to express my gratitude to the Center for Public Economics at the University of Dresden and my advisor Professor Dr. Marcel Thum.

Appendix A See Table A1.

Appendix B See Table B1.

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i

14

C.C. Corrigan / Resources Policy ∎ (∎∎∎∎) ∎∎∎–∎∎∎

References Aaronson, S.A., 2011. Limited partnership: business, government, civil society and the public in the Extractive Industries Transparency Initiative (EITI). Public Admin. Dev. 31 (S1), 50–63. Aguilar, J., Caspary, G., Seiler, V., 2011. Implementing EITI at the Subnational Level: Emerging Experience and Operational Framework. Extractive Industries for Development Series #23. World Bank. Alexeev, M., Conrad, R., 2005. The elusive curse of oil. Rev. Econ. Stat 91 (3), 586–598. Allison, P., 1999. Multiple Regression: A Primer. Pine Forge Press, Thousand Oaks, California. Andreula, N., Chong A., Guillén J., 2009. Institutional Quality and Fiscal Transparency. IDB Working Paper Series. 125. Bhattacharyya, S., Hodler R., 2009. Natural Resources, Democracy and Corruption. Working Papers Series Nr. 1047. University of Melbourne, Department of Economics. Boschini, A., Pettersson J., Roine J., 2005. Resource Curse or Not; A Question of Appropriability. Working Paper Series in Economics and Finance Nr. 534. Stockholm School of Economics. Busse, M., Gröning, S., 2013. The Resource Curse Revisited: Governance and Natural Resources. Public Choice 154, 1–20. Caspary, G., 2012. Practical Steps to Help Countries Overcome the Resource Curse: The Extractive Industries Transparency Initiative. Global Governance 18, 171–184. Collier, P., 1999. On the Economic Consequences of Civil War. Oxford EconomicPaper. 51(1), 168–183. Collier, P., 2002. Economic Causes of Civil Conflict and their Implications for Policy. Development Research Group, World Bank. Collier, P., 2006. Is Aid Oil? An Analysis of Whether Africa Can Absorb More Aid. World Development 34 (9), 1482–1497. Collier, P., 2007. The Bottom Billion: Why the Poorest Countries are Failing and What Can Be Done About It. Oxford University Press, Oxford. Collier, P., Hoeffler A., 2008. Beyond Greed and Grievance: Feasibility and Civil War. CSAE Working Paper Series, Nr. 2006-10. University of Oxford. Department of Peace and Conflict Research, Uppsala Universitet, 2010. UCDP/PRIO Armed Conflict Dataset v. 4-2010, 1946-2009. Uppsala, Sweden. Dreher, A., Voigt S., 2008. Does Membership in International Organizations Increase Governments' Credibility? Testing the Effects of Delegating Powers. CESIFO Working Paper, Nr. 2285. University of Munich. Extractive Industries Transparency Initiative, 2009a. History of EITI. http://eiti.org/ eiti/history, Accessed 8 June 2011. Extractive Industries Transparency Initiative, 2011. EITI Rules, 2011 Edition Including the Validation guide. S. Bartlett, K. Andreasen (Eds.). The EITI International Secretariat, Crown Copyright. Oslo, Norway. Extractive Industries Transparency Initiative, 2013. EITI Countries. EITI Secretariat, Oslo, Norway. Extractive Industries Transparency Initiative, 2013b. EITI Factsheet 2013. EITI Secretariat, Oslo, Norway. Feldt, H., 2009. Freiwillige Initiativen im Rohstoffsektor: Welche Ansätze bieten diese Initiativen zur Verbesserung der Governence in Rohstoffsektor in Entwicklungsländer?. Heinrich-Böll Stiftung, Berlin. Global Witness, 2009. Five Challenges for EITI to Deliver. http://www.globalwit ness.org, Accessed 1 August 2013. Hilson, G., Maconachie, R., 2009. Good Governance and the Extractive Industries in sub-Saharan Africa. Miner. Process. Extr. Metall. Rev. 30 (1), 52–100.

Iimi, A., 2007. Escaping from the Resource Curse: Evidence from Botswana and the Rest of the World. IMF Staff Papers. 54 (4). International Monetary Fund. International Monetary Fund, 2010. World Economic Outlook Database. www.imf. org/external/pubs/ft/weo/2010/02/weodata/index.aspx, Accessed 22 June 2011. Isham, J., Woolcock M., Pritchett L., Busby G.. 2003. The Variety of Resource Experience: How Natural Resource Export Structures Affect the Political Economy of Economic Growth. Middlebury College Economics Discussion Paper No. 03-08R. Islam, R., 2003. Do More Transparent Government Govern Better? Policy Research Working Paper, 3077. The World Bank, Poverty Reduction and Economic Management Division. Kaufmann, D., Kraay A., Mastruzzi M., 2010. The Worldwide Governance Indicators: Methodology and Analytical Issues. Policy Research Working Paper, Nr. 5430. The World Bank Development Research Group, Macroeconomic and Growth Team. Kolstad, I., Wigg, A., 2008. Is transparency the key to reducing corruption in resource-rich countries? World Dev. 37 (3), 521–532. Leite, C., Weidmann J., 1999. Does Mother Nature Corrupt? Natural Resources, Corruption, and Economic Growth. IMF Working paper 99/85. Marshall, M., Gurr, T., Jaggers, K., 2010. Polity IV Project: Political Regime Characteristics and Transitions, 1800-200. Center for Systematic Peace, Virginia, USA. McGuirk, E., 2010. The Illusory Leader: Natural Resources, Taxation and Accountability. Department of Economics, Trinity College, Dublin. Mehlum, H., Moene, K., Torvik, R., 2006. Institutions and the resource curse. Econ. J. 116, 1–20. Mejía Acosta, A., 2013. The impact and effectiveness of accountability and transparency initiatives: the governance of natural resources. Dev. Policy Rev. 31, 89–105. Ölcer, P., 2009. Extracting the Maximum from the EITI. OECD Development Center Working Paper, Nr. 276. OECD. Pitlik, H., Frank, B., Firchow, M., 2010. The demand for transparency: an empirical note. Rev. Int. Org. 5, 177–195. Robinson, J., Torvik, R., Verdier, T., 2006. Political foundations of the resource curse. J. Dev. Econ. 79, 447–468. Ross, M., 2001. Does Oil Hinder Democracy? World Polit. 53 (3), 326–361. Ross, M., 2002. Oil, Drugs and Diamonds: How Do Natural Resources Vary in their Impact on Civil War. Department of Political Science, UCLA. Sachs, J., Warner, A., 1997. Natural Resource Abundance and Economic Growth. Center for International Development and Harvard Institute for International Development. Harvard University, Cambridge, MA. Sachs, J.D., Warner, A.M., 2001. The curse of natural resources. Eur. Econ. Rev. 45 (4–6), 827–838. Sala-i-Martin, X., Subramanian A., 2003. Addressing the Natural Resource Curse: An Illustration from Nigeria. NBER Working Papers Nr. 9804. National Bureau of Economic Research, Cambridge, MA. Smith, S.M., Shepherd, D.D., Dorward, P.T., 2012. Perspectives on community representation with the extractive industries transparency initiative: experiences from South-East Madagascar. Res. Policy 37, 241–250. Stiglizt, J., 1999. On Liberty, the Right to Know, and Public Discourse: The Role of Transparency in Public Life. Oxford Amnesty Lecture, Oxford. The International Bank for Reconstruction and Development and the World Bank, 2011. World Development Indicators. Washington, D.C. United Nations Conference on Trade and Development, 2002. UNCTAD Statistics Overview. www.unctad.org, Accessed 8 June 2011. Yaffee, R., 2003. A Primer for Panel Data Analysis. New York University, www.nyu. edu/its/pubs/connect/fall03/yaffee_primer.html, Accessed 27 April 2011.

Please cite this article as: Corrigan, C.C., Breaking the resource curse: Transparency in the natural resource sector and the extractive industries transparency initiative. Resources Policy (2013), http://dx.doi.org/10.1016/j.resourpol.2013.10.003i