Transforming institutional quality in resource curse contexts: The Extractive Industries Transparency Initiative in Myanmar

Many resource-rich countries face the paradoxical situation that their wealth in natural resources coincides with low economic and human development rates. To address this so-called resource curse, academics and practitioners turn their hopes to institutional quality. Yet whether, how and with what consequences institutional quality is transformed in resource curse contexts remains poorly understood, especially so at subnational levels. The most widely implemented initiative that seeks to address the resource curse through enhanced institutional quality is the Extractive Industries Transparency Initiative (EITI). This article analyses to what extent and how the EITI transforms institutional quality at national and subnational levels in Myanmar, focusing on transpar- ency, civil society participation and accountability. We show that many transformations go beyond the o ﬃ cial EITI process and report. While the EITI report itself is not heavily used by civil society organisations (CSOs), the EITI process motivated CSOs to gather data and organise themselves both around and beyond EITI-related issues at subnational levels. Such participatory processes of constituting transparency improved relations between the (regional) government, CSOs and private companies, but also created new forms of in- and exclusion among civil society. While avenues opened up for CSOs to demand accountability regarding the impacts of resource ex- traction, the extent to which they are able to trigger action of extractive industry actors in their region remains limited. In conclusion, we argue that transformations in institutional quality are not characterised by a linear trajectory from transparency in the form of the EITI report to accountability, facilitated by civil society participation in EITI multi-stakeholder groups, as the EITI standard posits. Rather, transformations in institutional quality are characterised by spin-o ﬀ s, dynamic interlinkages, trade-o ﬀ s, limitations and a reinforcing cycle between participation and transparency within and beyond the EITI.


Introduction
Many resource-rich countries around the world face the paradoxical situation that their wealth in natural resources coincides with lower economic and human development rates than countries without resource abundance. This so-called 'resource curse' is characterised by high levels of corruption, domestic conflict, adverse environmental and social impacts of resource extraction, economic and socio-political inequalities, and an overall poor macroeconomic performance (Collier, 2017;Stevens and Dietsche, 2008).
To elucidate the conditions under which resource abundance translates into a resource curse, academics and practitioners have turned to analysing institutional quality (for an overview, see Papyrakis, 2017). Many argue that countries are less likely to experience the resource curse if they are able to effectively and democratically govern their natural resources (e.g. Boschini et al., 2013;Mehlum et al., 2006;Sarmidi et al., 2017). Though an increasing number of governance initiatives around the world is based on the assumption that improvements in institutional quality are important in addressing the resource curse, whether and how this works remains poorly understood. While some governance initiatives helped countries to alleviate symptoms or even causes of the resource curse (e.g. Caspary, 2012), others have been ineffective or even counterproductive in overcoming entrenched power asymmetries, opaqueness and resource conflicts (Phillips et al., 2016;Smith et al., 2012).
The most widely implemented governance initiative that seeks to reverse the resource curse through enhanced institutional quality is the Extractive Industries Transparency Initiative (EITI). The EITI is a global standard to increase transparency in countries' extractive industries sectors through the disclosure of information about licenses, production, revenues, payments, contracts and ownership (EITI, 2018a). The EITI is founded on the idea that transparency, in combination with civil society participation, can increase accountability in the extractive industries sector. So far, however, studies provide inconclusive evidence on the extent to which the EITI actually increases transparency, civil society participation and accountability, and whether it helps to address the resource curse (Corrigan, 2014;Papyrakis et al., 2017;Rustad et al., 2017). We argue that key to understanding the relation between institutional quality and the resource curse is a focus on the interlinkages between different aspects of institutional quality. This article advances such understandings by analysing to what extent and how the EITI transforms transparency, civil society participation and accountability in Myanmar, a country with prominent resource curse symptoms. While Myanmar has abundant natural resources, it is one of the least developed countries in the world, experiences severe adverse environmental and social impacts from resource extraction, and has a long history of domestic conflicts fuelled by the extraction of valuable resources (James, 2010;Perry, 2007;Pick and Thein, 2010). Myanmar became a candidate to the EITI in 2014, motivated explicitly by the aspiration to address its resource curse (Thein Sein, 2013;May Soe, 2014). The EITI also raised high expectations for boosting civil society's historically very low political participation and for addressing the adverse impacts from resource extraction. Myanmar is one of the few countries that has initiated EITI activities at the subnational level in order to stimulate wider public debates around natural resource governance. This makes Myanmar an interesting case to study to what extent and how the EITI transforms institutional quality in a resource curse context at the national and subnational levels.
This article draws on a literature review and twelve months of fieldwork in Myanmar by the lead author (September 2016-August 2017), during which most of the interviews were carried out to gain indepth understanding of the effects of the EITI. In June 2017, additional interviews were carried out by three of the authors in two regions in Myanmar. In total, the article draws on more than eighty semi-structured individual and group interviews in various regions of Myanmar, Chiang Mai (Thailand) and London with civil society, private sector and government representatives. Interviewees were selected based on their involvement in and knowledge of the design and implementation of the EITI in Myanmar.
We proceed as follows: the next section reviews literature on the relationship between the EITI, institutional quality and the resource curse, and identifies the EITI's theory of change. Section 3 describes Myanmar's natural resource context and governance, including the implementation of the EITI. Section 4 analyses to what extent and how the EITI in Myanmar transforms institutional quality at national and subnational levels. Section 5 revisits the EITI's theory of change, focusing specifically on the interlinkages between different aspects of institutional quality. Finally, Section 6 captures the article's main contributions and identifies areas for future research.
2. Addressing the resource curse by transforming institutional quality

The EITI's theory of change
In seeking to explain why some countries are able to translate their natural resource wealth into strong human and economic development while others experience the resource curse, the pivotal role of institutional quality has been widely recognised (Boschini et al., 2013;Brunnschweiler, 2008;Epremian et al., 2016;Mehlum et al., 2006;Stevens and Dietsche, 2008). The assumption is that high institutional quality can reduce (symptoms of) the resource curse by providing checks and balances against rent-seeking, corruption, excessive spending and an unequal division of the costs and benefits of resource extraction (Kolstad and Wiig, 2009).
The most widely implemented governance initiative that seeks to reverse the resource curse through enhanced institutional quality is the EITI, which is currently implemented in fifty-one (mostly developing) countries around the world. Countries that are candidate to the EITI are required to disclose information about their extractive industries sector through an official EITI report; establish a multi-stakeholder group (MSG) of government, private sector and civil society representatives to implement the EITI; and create public awareness and debates around natural resource governance (EITI, 2016).
The EITI was founded on the idea that "transparency and dialogue had to be part of the starting point" in addressing the resource curse (EITI, 2018b). This article assesses the validity of the EITI's theory of change, namely that "[t]he active participation of civil society in the EITI process is key to ensure that transparency created by the EITI leads to greater accountability" (EITI, 2016, 42). Through these changes, the EITI indirectly seeks to facilitate the "prudent use" of natural resources, contribute to sustainable development and prevent "negative economic and social impacts" from resource extraction, and thereby address the resource curse (EITI, 2016, 11) ( Fig. 1).

The role of transparency, participation and accountability
Studies exist-albeit inconclusive-on whether and how transparency, participation and accountability help to address the resource curse. Most studies focus on each of these institutional aspects separately, rather than on the interlinkages between them. In the following, we discuss causal assumptions between each of these aspects and the resource curse.
The idea behind enhancing transparency in resource curse contexts is that increased information access may help to constrain abilities of elites to appropriate resource rents and form patronage networks in opaque manners (Corrigan, 2014;Epremian et al., 2016;Kolstad and Wiig, 2009;Williams, 2011). Enhancing transparency may also increase trust and mitigate conflicts between stakeholders (Collier, 2017;Epremian et al., 2016;Haufler, 2010), including at subnational levels (Aguilar et al., 2011). However, transparency can only improve the governance of natural resources if affected actors can act upon the enhanced access to information. This ability depends on, among others, the quality of the available information, actors' abilities to process information, and actors' abilities to take or evoke action (Brockmyer, 2016;Epremian et al., 2016;Fox, 2007;Kolstad and Wiig, 2009;Stevens and Dietsche, 2008;Vijge, 2018).
Enhancing civil society participation in resource curse contexts can strengthen the representation or participation of affected stakeholders in the governance of natural resources. The EITI seeks to reorganise participation in resource governance by creating multi-stakeholder groups (MSGs). MSGs can create opportunities for civil society to raise their voice on unequal rent distribution or environmental and social impacts of resource extraction (Phillips et al., 2016). In many EITIimplementing countries, however, MSGs reveal unequal partnerships between stakeholders, a lack of full civil society participation and/or poor representation of affected citizens (Aaronson, 2011;Furstenberg, Transparency in the form of the EITI report

Accountability
Civil society participation in multi-stakeholder groups (MSGs) Addressing the resource curse EITI Fig. 1. The EITI's theory of change.
M.J. Vijge, et al. Resources Policy 61 (2019) 200-209 2015Haufler, 2010;Smith et al., 2012;Sovacool and Andrews, 2015). Finally, enhancing accountability of governments and extractive companies in resource curse contexts can help citizens to enforce responsibilities around, for example, the equitable sharing of revenues and the environmental and social impacts of resource extraction (Corrigan, 2014;Mejía Acosta, 2013). This requires citizens to be willing and able to call and effectively push for improvements in resource governance. Even if responsible actors are held to account and requested to take action (so-called answerability), they are not always willing or able to change their actions (Caspary, 2012;Collier, 2017;Epremian et al., 2016;Fox, 2007).

Does the EITI help to address the resource curse?
So far, studies show mixed results on whether the EITI actually reverses the resource curse through changes in transparency, participation and accountability.
On the one hand, Rustad et al. (2017) argue that the EITI has effectively created new standards for auditing and reporting in resourcerich countries and that EITI MSGs enhance civil society involvement in resource governance. Similarly, Caspary (2012) argues that the EITI enhances transparency and that EITI MSGs help build trust and collaboration between stakeholders. Also Corrigan (2014) points to some early signs of success in the EITI's ability to strengthen governance and address certain (though not all) aspects of the resource curse.
On the other hand, a number of studies find no difference before and during EITI compliance, and between EITI member and non-member countries in most indicators for the resource curse (Öge, 2017;Scanteam, 2011;Sovacool et al., 2016;Kasekende et al., 2016). Hilson and Maconachie (2009) argue that the EITI can only promote accountability if there are already strong institutions in place. Shaxson (2009, 45), even argues that the EITI draws attention away from "tougher issues related to the resource curse". Often-mentioned challenges for the EITI's effectiveness include limited awareness and knowledge on the EITI; different understandings of the EITI among stakeholders; and insufficient participation, which in turn leads to limited ways to hold decision-makers accountable (Aaronson, 2011;Caspary, 2012;Epremian et al., 2016;Rustad et al., 2017;Scanteam, 2011).
The above shows that the extent to which governance initiatives-and the EITI in particular-trigger changes in institutional quality that help to address the resource curse depends largely on contextual factors (Epremian et al., 2016). If so, reversing the resource curse in a context such as Myanmar, which recently embarked on a democratic transition after decades of military dictatorship, poses significant challenges. Given that transparency, participation and accountability are often seen as core elements of democracy, we consider it a crucial endeavour to analyse contemporaneous changes in these aspects during Myanmar's democratic transition.

The resource curse and the implementation of the EITI in Myanmar
Myanmar engages in offshore and onshore oil and gas extraction, as well as extraction of large amounts of gemstones and minerals. Both large-and small-scale mining takes place in many of Myanmar's fourteen states and regions (Fig. 2). While Sovacool (2010) points to shifting evidence for Myanmar's resource curse, in absolute terms the country exemplifies many resource curse symptoms (James, 2010;Perry, 2007;Pick and Thein, 2010;Thein Sein, 2013). Despite its wealth in natural resources, Myanmar remains one of the least developed countries in the world. The distribution of benefits from extractive industries is severely skewed. While Myanmar's extractive industries sector generates billions of dollars, most citizens do not profit from this (Global Witness, 2015). Adverse impacts caused by extractive industries include forced relocation, environmental pollution, destruction of livelihoods and negative health effects (Adam Smith International, 2015;Bauer et al., 2016). Myanmar also has longstanding and violent conflicts, many of which are fuelled and sustained by resource extraction (Holliday, 2012).
Myanmar's natural resource governance framework largely fails to address those issues. Environmental laws were formulated as recently as 2012 and remain poorly enforced (Adam Smith International, 2015). The governance of Myanmar's extractive industries sector is characterised by high levels of corruption and secrecy. According to the Resource Governance Index, which evaluates countries' extractive industries sector on transparency and accountability, Myanmar scored 83rd of the 89 countries (Natural Resource Governance Institute, 2017). A substantial proportion of Myanmar's resource extraction (for gemstones even an estimated 70-80%) is illegal and thereby fails to contribute to the state coffers (Irwin, 2016;Global Witness, 2015). Myanmar citizens have very limited opportunities to hold extractive industry actors to account for their actions. Grievances associated with resource extraction have historically been addressed with force rather than through dialogue (Adam Smith International, 2015).
Myanmar recently embarked on a major democratic and economic reform process. The country made a quick transition from a military regime to a quasi-civilian government in 2011, to a democratically elected government at the start of 2016. This reform process coincides with major adjustments in the country's natural resource governance framework, which include a number of new and amended laws and regulations on environmental and social impact mitigation for investment projects. Civil society participation has been actively promoted in these efforts (Kramer, 2011;Simpson, 2013). While some CSOs still operate from the borders of Myanmar, national and local civil society has become quite vibrant (Matelski, 2016), including around extractive industries issues.
As a "central part" of its democratic transition, Myanmar committed to implementing the EITI in 2013 (EITI, 2015). The EITI is seen as "a useful tool to design [Myanmar's] escape from the resource curse", as argued by the former national EITI coordinator (May Soe, 2014). Initially, Myanmar's history of civic oppression and human rights violations made it highly challenging for the three stakeholder groups-government, civil society and private sector representatives-to come together to discuss transparency. Despite this, the MSG managed to make quick progress, resulting in the publication of Myanmar's first EITI report in January 2016 (Vijge, 2018).
In 2015, the national MSG decided to establish EITI subnational coordination units (SNCUs) in Myanmar. These units were established to further stimulate awareness and discussion around the EITI and natural resource governance within MSGs at the subnational level (Vijge and Simpson, Forthcoming). Two states, Shan and Rakhine, and two regions, Magway and Mandalay, were initially chosen to host these units, mainly because of their abundant natural resources and a gas pipeline that runs through these areas, with reportedly large negative impacts. Due to challenges related to the vast size of Shan state and the conflicts in Rakhine state, the subnational units were only established in the regions Magway and Mandalay. 1 Magway has large reserves of coal, oil and gas, while Mandalay harbours gemstones, gold and other minerals (Fig. 2). The subnational MSGs that comprise the SNCUs in Magway and Mandalay include regional and/or local government, civil society and private sector representatives. These MSGs discuss and exchange information about region-specific issues related to resource extraction.
After the governmental transition at the start of 2016, the national EITI process was stalled for more than a year, thereby also putting a halt to the SNCUs in Magway and Mandalay. After Myanmar's successful application for a one-year extension, Myanmar reinitiated its EITI process in April 2017, and published its second and third EITI report in March 2018. Soon after reinitiating the EITI process, the national MSG decided to (re-)establish the SNCUs, though this decision has not yet been implemented at the time of writing. Due to the timeframe of the fieldwork, the article does not consider EITI developments or reports published after August 2017.

The EITI: transforming institutional quality in Myanmar?
This section analyses to what extent and how the EITI transforms institutional quality at national and subnational levels in Myanmar. A summary of the findings can be found in Table 1.

Transparency
The EITI aims to enhance transparency through the publication and dissemination of EITI reports. The publication of Myanmar's first EITI report in 2016 was widely heralded as a major achievement in shedding light on the largely opaque extractive industries sector. However, the quality of the data has been drawn into question due, in part, to the very short time-frame for consolidating data. 2 Also the uptake of the report's information by CSOs is rather limited. 3 An important reason for this is the lack of (region-)specific information, for example about the regional allocation of licenses, contracts, and the beneficial owners of extractive industries (Vijge, 2018). More important than transparency arising from the EITI report was the generation of information by stakeholders involved in EITI activities, information that went beyond what was disclosed in the EITI report. In the initial stages of Myanmar's EITI process, it was time-consuming and difficult for CSOs to request region-specific information from the government, since such requests had to be submitted to the union level. To obtain information beyond what was disclosed through the EITI report, CSOs involved in the national EITI process proposed the establishment of the above-mentioned subnational coordination units (SNCUs). 4 CSOs in both Magway and Mandalay claim that these units facilitated the exchange of information among government, civil society and private sector representatives. In some (though not all) instances, for example, the government responds to calls from CSOs to inspect extractive industry practices and engages CSOs in these inspections. 5 This information exchange, which goes well beyond the EITI report, is considered important to establish common ground for debates, thereby building understanding and mutual respect between stakeholder groups. Initially, for example, CSOs' interaction with companies was characterised as highly "emotional", with CSOs often drawing on assumptions in levelling blame against companies for negative impacts on surrounding communities. Gradually, CSOs learned to more effectively communicate with companies and engage in, as one interviewee called it, "data-based discussions". 6 This increased the quality of civil society participation in the discussions around resource extraction (see Section 4.2). Besides informing CSOs, the SNCUs have also enhanced companies' awareness and knowledge on their environmental and social impacts, on impact mitigation rules and measures, and on (inter) national business standards. 7 The SNCUs in Mandalay and Magway also triggered CSOs to actively collect data beyond what is being disclosed through the EITI report, sometimes with the help of local communities and/or in collaboration with the government and private sector. 8 In Mandalay, a so-called Mandalay EITI working group monitors mining activities in the 18 townships where adverse impacts of resource extraction are most acutely experienced. Drawing on information disclosed by the government and through the EITI report, the working group checks whether companies' mining licenses are still up-to-date and accurate for the area in which they operate. As one interviewee explained: "For example, in Pyaw Bwal Twin village, a lead factory was built (…). The villagers did not want that factory. They came to us [the Mandalay EITI working group] and we went to discuss (…). We found out that the factory (…) did not have permission to operate. We informed this to (…) the government, and then the regional government stopped the factory operation." 9 All this was triggered by, but happened outside the SNCUs, which were suspended in mid-2016.
Though the SNCUs in Magway and Mandalay have been an important trigger for the exchange and generation of information on (impacts from) the regions' extractive industries, CSOs in Magway and Mandalay are not fully satisfied with the information exchange between the three stakeholder groups. The regional government and companies in Magway are criticised for being slow and hesitant to disclose certain, often sensitive, information. For example, CSOs have in vain requested information on the beneficial ownership of military-or foreign-owned companies, important information that is nevertheless beyond what was disclosed through the EITI report. 10 Many civil society representatives are of the opinion that the EITI has not (yet) been effective in addressing deeply-engrained corruption within the government, crony companies, or military-or state-owned companies. 11 One might argue that the EITI has so far mainly reaped the low-hanging fruits. Illegal resource extraction still takes place on a large scale and remains unscrutinised in the EITI report. Furthermore,

Table 1
Transformations in institutional quality triggered by the EITI at national and subnational levels in Myanmar (own compilation).

Aspect of institutional quality Transformations
Remaining challenges Transparency EITI report sheds light on extractive industries sector Validity of data drawn into question. EITI report covers least problematic resource extraction. EITI has not (yet) addressed deeply engrained corruption Triggered by SNCUs, CSOs gather, exchange and use information beyond EITI report, including on impacts of resource extraction in the region Limited use of EITI report by CSOs, partly due to lack of (region-)specific information the selection of the four initially planned SNCUs excluded those regions where resource extraction is strongly linked to conflicts between the state army and ethnic armed groups, such as Kachin state. 12 It was only in mid-2017 that the national MSG decided to establish an SNCU in Kachin state. It remains to be seen whether, when and with what effects this unit will start to operate.

Participation
As argued in Section 2, enhanced civil society participation in decision-making may improve natural resource governance in resource curse contexts by creating opportunities for affected stakeholders to be represented and/or raise their voice. The multi-stakeholder negotiations that preceded Myanmar's EITI candidature in 2013 were the first ever opportunities for CSOs to enter into political discussions with the government. CSOs generally state satisfaction with their involvement in the national EITI process (Vijge, 2018). After decades of civic oppression, it was unprecedented for Myanmar that-unlike in some other EITI-implementing countries (see e.g. Aaronson, 2011;Kolstad and Wiig, 2009)-civil society participation was not obstructed, but instead promoted by the government (Vijge and Simpson, Forthcoming).
Similarly, the SNCUs in Magway and Mandalay opened up political space for CSOs to engage in discussions around natural resource governance: "Because of the EITI, people have a voice". 13 Through their engagement in the SNCUs, CSOs significantly built their capacity to lobby and collaborate with the government and private sector. 14 CSOs gained tools and venues, as well as a sense of safety that they had long lacked. Most importantly, as formulated by a CSO representative: "[t]he best outcome we got was respect and trust (…) [among] the groups." 15 This "mutual understanding" 16 between government, civil society and private sector representatives provided CSOs with better access to certain, though not all, government agencies and companies. This allowed them to voice their opinion regarding resource extraction in their region. 17 CSOs felt that this was particularly relevant in discussions around the environmental and social impacts in Magway and Mandalay; issues that, again, fall outside the scope of the EITI. 18 While the subnational discussions around natural resource governance happened as a result of the national EITI process, the resulting transformations go well beyond its confines. After the national EITI process was suspended at the beginning of 2016, national-level support for the SNCUs discontinued, but civil society participation in subnational resource governance persisted. In Magway, for example, CSOs have taken a leading role in initiating EITI-and natural resource-related activities through a so-called EITI watch group. Although the EITI was the motive to establish the EITI watch group, its activities consist of organising discussions with companies and joint monitoring of the environmental and social impacts of resource extraction in the region. 19 It is important to note that subnational discussions and activities around such adverse impacts happen despite the fact that the national MSG decided against including these issues in the EITI report. CSOs exerted large pressure and even threatened to withdraw from the national EITI process if impacts could not be discussed. The government's aspiration to make quick progress towards compliance with the more narrow EITI requirements was ultimately decisive in the resolution to exclude environmental and social impacts from the EITI report (Vijge, 2018). Ostensibly, trade-offs exist at the national level between (quick) compliance with the EITI standard and inclusiveness of voice, as one interviewee also asserted: "We need the right balance. Yes, the process needs to be successful, but it also needs to be meaningful so that it has traction with broader citizens and lead to meaningful reform and changes for the lives of people". 20 These trade-offs are less prominent at the subnational level, since SNCUs are not a requirement of the EITI standard.
While the above shows a high level of civil society participation in and beyond the SNCUs, the prominent role of civil society is also linked to limited engagement of other actors in the official EITI process. Even before the suspension of the national EITI process, for example, CSOs criticised the regional governments of Magway and Mandalay for their limited understanding of and engagement in the SNCUs. 21 Furthermore, since participation in the SNCUs is voluntary, companies with large adverse environmental and social impacts are generally not involved in the SNCUs. 22 To organise civil society representation in the national and subnational EITI process, a core group of CSOs established the Myanmar Alliance for Transparency and Accountability (MATA) in 2014. MATA is one of the first CSO alliances in Myanmar, consisting of reportedly 580 CSOs. 23 MATA is founded on the idea that civil society needs "to speak with one loud voice" 24 in the EITI process in order to form a strong negotiating block, particularly vis-à-vis the historically very strong government and military. 25 The establishment and functioning of MATA are a clear indication of enhanced civil society participation and empowerment resulting from the EITI process: "MATA helps to build strength and better connect to people". 26 The perceived need to speak with one voice, however, has also created disputes among CSOs. Some CSOs are critical of the fact that only MATA members represent civil society in the EITI process: "there is not a variety of voices, just the voice of MATA". 27 Some claim that ethnic and border-based organisations are "marginalised" within MATA, and that "there is a Yangon-centralisation. Funding-oriented. It's more a parent-child relationship between Yangon-based groups and subnational groups". 28 Some CSOs also have doubts about the extent to which MATA represents communities that experience adverse impacts of resource extraction. 29 They argue that through their involvement in the EITI process, MATA members "lose touch with the people on the ground". 30 Other CSOs criticise so-called "dirty CSOs" for being closer to the government and private sector than to citizens 31 (see also Vijge, 2018). Still other CSOs, mainly those that are border-based, deliberately choose not to engage in the EITI process, arguing that it "is regulated by Yangon and Nay Pyi Taw" and that it "takes the attention away from the real issues". 32 Given these new forms of in-and exclusion in civil society participation, some go so far as to argue that the EITI process divided rather than united civil society. 33 CSOs indicated their concern that due to these disputes, regional and national government officials have "less trust in CSOs" 34 and their ability to represent Myanmar citizens. On the other hand, the debates about representation also indicate an increasing political space for, and self-confidence of, diverse civil society actors to raise their specific concerns in political contestations.

Accountability
The EITI is based on the assumption that enhanced transparency and participation in the extractive industries sector help citizens to hold responsible actors to account for the management of natural resources. Since the EITI report in Myanmar has not been heavily taken up by CSOs, its effects on accountability in the extractive industries sector are limited. However, the EITI process, including the SNCUs, did trigger transformations in accountability, though to a limited extent.
An important transformation is the enhanced ability of CSOs and citizens to demand action of the government and extractive industry companies. Because of the MSG discussions that were held as part of the SNCUs, CSOs can demand accountability if they think the government or companies fail to follow EITI standards or other rules and regulations, for example related to corporate social responsibility (CSR) and tax payments and use. 35 As one interviewee argued: "They [companies] are doing CSR because in the MSG, the CSOs are asking what kind of activities they do in CSR". 36 In the subnational MSG discussions, CSOs will also be "pushing companies to pay taxes and [pushing] the government for the taxes to be beneficial for the people and for the country (…) according to the EITI standards." 37 EITI-related activities also enhance communities' and their leaders' capacities to assess the advantages and disadvantages of extractive industry investments in their region and their rights in demanding impact mitigation or benefitsharing. 38 This again shows that the transformations in transparency and accountability go well beyond the EITI mandate.
Though the EITI process made it possible for CSOs to demand accountability on resource extraction, the extent to which CSOs are also able to evoke action by responsible actors is limited. Various interviewees indicated that the SNCUs have so far failed to change the behaviour of extractive industry actors: "Many meetings, but very little effects". 39 At least one prominent CSO discontinued its engagement in the EITI process, claiming that there had not been any changes in resource extraction on the ground. 40 Indeed, though many actors involved in Myanmar's EITI process strongly believe that the EITI has the potential to reduce adverse impacts from resource extraction, 41 many also acknowledged that the impact of the EITI on actual practices of resource extraction remains limited to date. 42 There are substantial challenges to enhancing accountability in Myanmar's extractive industries sector. The capacity and knowledge base of citizens and civil society on natural resource governance in Myanmar is still very low, particularly on highly technical topics such as the EITI: "We don't use [the EITI] because we don't know what the EITI is". 43 Furthermore, institutional mechanisms to hold actors to 25 Authors' interviews with former MATA Mandalay representative, 4-6-2017; account, such as project-level grievance mechanisms, are rarely available or accessible to Myanmar citizens or CSOs. 44 The fact that CSOs use the EITI process well beyond its official mandate also shows the perceived lack of more appropriate venues to hold actors to account on issues related to resource extraction.
Another challenge to enhancing accountability through the SNCUs is the limited decentralisation in Myanmar. State and regional governments have limited mandates to manage their natural resources or handle and spend taxes collected in their region (Adam Smith International, 2015;Lynn and Oye, 2014). This reduces CSOs' ability to evoke actions from their regional government. It also limits the relevance of subnational MSGs. As one interviewee argued: "We only have meetings to discuss. We don't have meetings to solve the problems." 45 In a number of instances in Magway and Mandalay, CSOs submitted reports regarding the impacts of extractive industries in their region to the union level government. 46 Most interviewees making reference to this, however, said that they did not receive any response, let alone witness any changes on the ground. 47

Discussion: revisiting the EITI's theory of change
This section revisits the EITI's theory of change, presented in Section 2: "[t]he active participation of civil society in the EITI process is key to ensure that transparency created by the EITI leads to greater accountability" (EITI, 2016, 42). Our results are depicted in Fig. 3.
As we showed, the EITI in Myanmar has offered unprecedented opportunities for CSOs to raise their voice on the governance of the extractive industries sector. As a result of enhanced civil society participation in the EITI process, CSOs learned to interact and collaborate with the government and private sector, which contributed to the development of common ground and understanding between these groups. Beyond the outline of the EITI's theory of change, we showed that such participation and collaboration stretch across the boundaries of the official MSG meetings. Here, interlinkages between the different aspects of institutional quality become apparent: enhanced transparency in the form of self-generated information enabled CSOs to more effectively communicate with companies, debate social and environmental problems of mining, and, as a result, demand accountability, even on issues that are beyond the EITI. This not only means that civil society participation in resource governance facilitates the use of disclosed information to hold actors to account-as posited in the EITI's theory of change-but also that participation is facilitated by the increased availability of information. As such, the EITI process triggers a self-reinforcing cycle in which participation leads to enhanced transparency, which in turn further facilitates participation (circular arrow in Fig. 3).
To organise civil society representation in the EITI process, CSOs established the civil society alliance MATA. This resulted in significant capacity-building and representation of some, but not all, CSOs. These new forms of in-and exclusion triggered conflicts among CSOs about rightful representation. Arguably, trade-offs exist between, on the one hand, CSOs' perceived need to speak with one voice and, on the other hand, maintaining representation and on-the-ground understandings of a diversity of actors affected by resource extraction. This reveals the inherent limitations of civil society participation, even in a case where there is no evidence for active obstruction by the government, which is often identified as the main bottleneck for participation in other EITIimplementing countries (Aaronson, 2011;Kolstad and Wiig, 2009;Öge, 2017;Ospanova et al., 2013;Wilson and Alstine, 2014).
In early 2016, the official EITI process was stalled for more than a year. Unexpectedly, however, relations that had been built or improved in the SNCUs were reorganised in spin-off meetings and activities, allowing CSOs to continue to demand, exchange and generate (region-) specific information related to resource extraction. This proved relevant in holding extractive industry actors to account for their social and environmental impacts, issues that go well beyond the scope of the EITI. As such, accountability was furthered not so much by the "transparency created by the EITI" (EITI, 2016, 42), but particularly by the generation and exchange of information that goes beyond the EITI.
Though the EITI report and process enhanced transparency in Myanmar's extractive industries sector, the EITI has so far mainly reaped the low-hanging fruits. The EITI has not yet shed light on illegal resource extraction, corrupt practices by state and non-state actors, or resource extraction that is tied up with ethnic conflicts. There are also important limitations in the extent to which the EITI in Magway and Mandalay enhanced accountability. Though CSOs' ability to demand accountability has been enhanced, scarcity of accountability mechanisms and limited decentralisation in natural resource governance constrain CSOs' ability to evoke actions from extractive industry actors in their region (see also Vijge and Simpson, Forthcoming). In other words, while the EITI has generated some answerability (Fox, 2007), the transparency and participation it has triggered are not sufficient to generate full accountability (represented by the dotted lines in Fig. 3).

Conclusion
Empirically, this article showed that-despite the limitations-the EITI can trigger important transformations in institutional quality at national and subnational levels in a resource curse context like Myanmar's. Theoretically, we conclude that transformations in institutional quality are not necessarily characterised by a linear trajectory of enhanced transparency-in the form of the EITI report-leading to accountability, facilitated by civil society participation in multi-stakeholder groups (EITI, 2016). Instead, transformations are characterised by spin-off effects, dynamic interlinkages, reinforcing cycles, and emerging trade-offs and limitations of different aspects of institutional quality (Fig. 3). We argue that such complexity should be taken Transparency in and beyond the EITI report Answerability, but not full accountability Civil society participation in and beyond MSGs EITI Fig. 3. Revisiting the EITI's theory of change: the spin-off effects, reinforcing cycle (↻) and limitations (represented by dotted arrows) between transparency, participation and accountability.
into account in the implementation of transparency initiatives such as the EITI in resource curse contexts. Methodologically, we argue that analyses that acknowledge this complexity are better able to contribute insights into how institutional quality may address the resource curse than analyses that draw static analytical boundaries around institutional quality as a single unit of study.
Also important in such analyses is consideration of how governance initiatives such as the EITI interact with broader transformation processes. In Myanmar, the EITI is seen as a "central part" of the country's democratisation process, which also coincides with multiple amendments in the natural resource governance framework. This gives rise to analytical challenges in determining to what extent the EITI triggers or merely coincides with broader-scale processes. In our case, while transformations in institutional quality may also be considered a result of broader political transformations, interviewees explicitly referred to the EITI as key in triggering these changes and in, as some framed it, "improv[ing] our country towards a democratic society". 48 It is still early days for many of the amendments in Myanmar's natural resource governance framework to generate outcomes on the ground. Subsequent analyses in Myanmar as well as other countries would therefore be welcome to assess how the EITI transforms (interlinkages between) aspects of institutional quality through time, how such changes interact with broader reform processes, and how these translate into substantive changes to address the resource curse.