Development aid and domestic regional inequality: the case of Myanmar

ABSTRACT

There is a cyclical nature to the dilemmas confronting international donors willing to operate in Myanmar. Brief periods of relative openness led to rapid surges in development assistance, regularly interrupted by long phases of military rule and disengagement by donors. Amidst all this, many predicaments remain. This article engages with one of them: the inequality between regions. How have international donors reacted to the issue of domestic regional inequality? Recent studies suggest that official development assistance (ODA) does not target poor regions very well, but it is not always clear why this is the case. Myanmar’s sudden, yet uneven and unequal liberalization from 2011 to 2021 catalyzed huge inflows of ODA, while it also confronted donors with new policy dilemmas. The article shows that aid providers struggle with the problem of rising regional inequality, especially for political reasons. Donor and recipient interests often do not align well on this issue. In the case of Myanmar, donors who press for regional inequality to sit prominently on the agenda might fare less successfully than those who address the issue indirectly. The article concludes that regional inequality and the politics of targeting deserve a more central role in the political economy of ODA.

KEYWORDS:

• Myanmar (Burma)
• development assistance
• regional inequality
• knowledge production
• positionality
• aid targeting

Introduction

Inequality between the regions of a country (“domestic regional inequality”) can have tremendous political consequences. It reinforces and exacerbates divides along “urban” vs “rural” or “central” and “peripheral” lines, especially in developing countries. Regional divides can be regarded as “meta-divides” since they often contain differences in income, culture, and social structures. Scholars have shown that regional inequality has significant implications for grievances, conflicts, and political contestation (Alesina et al. 2016; Cederman et al. 2011; Østby et al., 2009) and where there are conflicts over territory and the exploitation of resources (Bates 1983; Berkowitz 1997).

Against this background, the role of Official Development Assistance (ODA)¹ in mitigating or exacerbating regional inequality is a topic of growing importance. While traditionally the academic literature has focused on the relationship between ODA and growth and poverty (Sumner and Glennie 2015), in recent years more fine-grained data have allowed scholars to delve into the domestic distribution of ODA (Bluhm et al. 2018; Briggs 2017; Nunnenkamp et al., 2017). One finding of this literature is that for many countries ODA tends to flow to relatively richer regions and hence shows problems in targeting poorer ones (Manuel et al. 2019). Overall, a better understanding of when and why aid agencies struggle with addressing regional inequality is necessary.

In this article, we use Myanmar² as a case study to investigate these questions. Myanmar is a particularly instructive case because of the fast political changes between 2011 and 2021 and their implications for the aid sector. Although generally presented as a “transition”, the process was contested, uneven, unequal, and so was its labeling (Stokke and Soe Myint Aung 2020).³ An assessment of what happened needs to be situated as this was deeply racialized and gendered, with hierarchies and violent exclusions apparent even within the dominant Buddhist-Bamar core (Campbell and Prasse-Freeman 2022; Stokke and Soe Myint Aung 2020; Hedström and Olivius, 2022). This “transition” triggered both a dramatic socio-economic transformation and a substantial inflow of ODA as sanctions were lifted and debt forgiven (Bissinger 2017; Holliday and Zaw Htet 2017; Fumagalli 2022b). Reversing the relatively long and strong isolation from the world markets created a time-lapsed version of processes that are usually much slower and more difficult to observe. Myanmar is also a notoriously difficult case for rolling out aid to the periphery, with its long heritage of resisting centralization (Carr 2018; International Crisis Group [ICG] 2002, 2004; Meehan and Sadan 2018; Scott 2009), its decades-long ethnic armed insurgencies (Callahan 2003; Parks et al. 2013; South 2008) and a model of “ceasefire capitalism” modeled around resource extraction (Hardaker 2020; Kramer 2021; Woods 2011, 2019).⁴

Our findings contribute to the existing literature on the domestic distribution of ODA and the politics of aid targeting. While most studies focus on project aid, we also look at the overall balance of aid flows, as project aid is often dwarfed by other forms of assistance. We also explore the ways in which major donors interact with the recipient government. Given that it is the main gateway for aid in a contested political territory, international donors need the consent of the recipient government.⁵ Thus, explicit forms of targeting might backfire if they were to be perceived by the central authorities as politically motivated. In our research we draw on quantitative and qualitative methods to explore both the evolution of aid flows to the country, paying attention to both traditional and new donors (Japan, South Korea, the United Kingdom and Germany). We complement this analysis with a case study on higher education reform lead by a non-state donor in which we were also involved and about which we conducted interviews between 2013 and 2022.

The article’s main contention is that international donors struggle with regional divergence not only for logistical but also for political reasons. Donor and recipient interests often do not align well on this issue. In the case of Myanmar, donors who press for regional inequality to sit prominently on the agenda might fare less successfully than those who address the issue indirectly. While donors alone hardly account for the rising inequality, the substantial aggregate aid flows do little to mitigate it. Only in recent years have we seen efforts to counter this imbalance. One of the problems lies with conflicting donor motives. While most donors now have rural development as one of their priority areas, rolling aid out to the “periphery” remains challenging and not always advantageous for these regions in practical as well as political terms (Carr 2018; International Crisis Group (ICG), 2002). The second and related reason lies in the recipient’s interests: in the context of a highly centralized government, large portions of international aid gravitate toward central regions. Where donors explicitly try to target others, they will face resistance from the recipient’s side.

The period we focus on in this article covers the decade of sudden and short-lived and partial political opening between 2011 and 2021. We will not cover the period from 1 February 2021 onwards, when a coup by the Myanmar military brought an end to the changes introduced over the previous decade (Fumagalli 2022a; Loong 2021). The coup itself might have been partly – among other things – the consequence of regional polarization and contestation, but it represents a clear break as most western donors have stopped providing ODA to Myanmar.

The article proceeds as follows. First, we discuss our methodology and positionality in doing research in Myanmar. Next, we survey the scholarly literature on regional inequality and the role of ODA. Subsequently, we look at the Myanmar case and discuss important contextual factors. In the remainder of the article we examine the quantitative evidence for the relationship between aid flows and regional inequality, before delving in greater depth into the examples of a select number of donors. The final section concludes with some general implications of our findings.

Positionality, methods, and data

Research in authoritarian and conflict settings confronts scholars with a particular set of challenges and responsibilities. In our work we are followed best practices as discussed in anthropology (Chattopadhyay 2013), sociology (Wackenhut 2018) and geography (England 1994; Koch, 2013), as well as political science (Knott 2019; Malejacq and Mukhopadhyay 2016; Fumagalli and Rymarenko 2022; Fumagalli 2023). The Burmese Studies scholarship has been particularly attentive to issues of decolonization and neo-coloniality in the study of Burma/Myanmar, including in relation to sources, methods and collaboration with local scholars (Brooten and Metro 2014; Metro 2019; Tharaphi Than 2021; Chu Main Paing and Than Toe Aung, 2021; Hedström 2019; Décobert 2014), cautioning against “white saviourism” (Calabrese and Cao, 2021; Maung Zarni, 2021) and neo-coloniality in research (Chu May Paing and Than Soe Aung 2021; Phyo Win Latt 2011; Tharaphi Than 2021). This important strand of the literature also advised against a casual engagement with local authors, scholarship and sources (Charney, 2021a, 2021b).

As white male scholars from Global North institutions, we are aware that our position in Myanmar came with considerable privilege. Privilege, “a matter of social power and political differentiation” (Kappler, 2021, 422), is a condition that international researchers – especially white, especially from the Global North – often enjoy (McCorkel and Myers 2003). For a start, entry and exit were reasonably unproblematic for us, as we could rely on the support of local academic institutions and the Ministry of Education for relevant visa support. This was the case both during the Thein Sein administration (our visits began in 2013) and during the NLD government (our last visit was in 2019). Exit or even intra-country mobility were privileges not afforded to our local academic colleagues, who had to apply to the Ministry of Foreign Affairs to release their passports for international visits to our former institution in Europe (which had a large institutional agreement with Myanmar’s universities). Another issue was that of the status attached to our role. Although our research on international aid was supported by other institutions, especially during the 2015–2022 period, our initial access to the country and engagement with local universities was facilitated, indeed enabled, by a large non-state donor, which from 2013 onwards became a major source of support for higher education reform in the country. This carried a high risk of being identified as donors ourselves or being able to influence the donor’s funding decisions (both assumptions were incorrect). We were very clear at all times that this was not the case, being conscious of the serious and unethical risk of mis-identification (Tewksbury and Gagné 1997). We see our own stance as one of “positioned interpretation” (Mosse 2006, 941 and Décobert 2020). We position ourselves as part of the project we designed and helped implement and are responsible for its contributions and its failings.

To examine how aid agencies address the issue of regional inequality, we draw on three different empirical sources. Quantitatively, we compiled a data set on aid flows to Myanmar’s regions,⁶ geocoding ODA flows regionally over some 20 years. We briefly compare these data with the data on regional inequality based on satellite images to see whether aid flows have responded to the skyrocketing levels of regional inequality in recent years (see Appendix C). Next, we review the rhetoric and practices of four other bilateral donors (Germany, South Korea, Japan and the UK) to understand the degree to which donors respond to regional inequality and whether different donors respond differently. We decided not to include two other major bilateral donors (US, Australia) as they are somewhere in the middle of the pack when we look at regional shares, giving clearly more to regions than a donor like South Korea, and decidedly less compared to Japan or Germany (see Appendix A, Table A1 and Table A2 below). Consequently, they would not add much variation to our comparison. An exception might be China, but it has been extensively discussed in the literature including in relation to Myanmar (Reilly 2013; Steinberg and Fan 2012; Kiik, 2016; Rippa 2020; Calabrese and Cao, 2021). Data on Chinese aid flows are also challenging to compare to those of western donors. Third, we also rely on our own observations and interviews carried out during multiple visits to Myanmar between 2013 and 2019 for periods between two and eight weeks a year, to highlight the limits of donor agency and to tease out the role and agenda of the Myanmar government. Additional visits were made by Fumagalli to Bangladesh (Cox’s Bazar) in 2020 and Thailand (Chiang Mai) in 2022, as he has been involved in research on refugees, conflict and cross-border aid in the educational sector on both sides of the Myanmar border. Altogether, we conducted 24 interviews with representatives of international donors, local academics and experts (see Appendix B).

An additional note on terminology and data is also in order, as we see honesty and transparency with data limitations as crucial to the process of knowledge production. While in our paper we appear to rely on “hard (statistical) data” also, we do so with the awareness that data are never neutral nor speak for themselves. This is nowhere more evident than in Myanmar. Our units of analysis are the administrative units in Myanmar (seven States, seven Regions, and the capital Nay Pyi Taw). The challenge in this regard is thus two-fold: one, getting “precise numbers”, because the Myanmar government does not reveal census data on ethnicity in each state and division. And two, problematizing those numbers given that ethnonyms such as “Karen” mean different things to different people at different times in different locations (Décobert 2020; Décobert and Wells 2020; on other groups and the historical context behind classification see Ikeya 2020 and Candier 2019). The same point of course applies to the case of the Kachins (Sadan 2013), as there are Kachins outside of Myanmar’s Kachin State, and outside Myanmar for that matter. Furthermore, ethnic Bamars live in Kachin state too, especially in the relatively larger urban centers (Sadan 2013). The Kachin war itself is less about Kachin State as such, as much fighting takes place in Northern Shan State. Last, but not least, the terms “nation” and, concomitantly, “subnational” to refer to domestic regional inequality, while common in the aid literature, are controversial and contested in the Myanmar context. It is therefore important to use terms such as “periphery” with extreme caution. Our intention is not to reify politically loaded concepts in a geographic or ethnic dimension. To the contrary, the idea of a “Bamar centre” roughly consistent with the Irrawaddy basin and a “non-Bamar periphery” is itself a historical construct (Charney 2009, especially chapters 1&2) to a large extent created by British colonialism and then perpetuated after independence.⁷

The problem of regional inequality and the role of ODA

The regional dispersion and distribution of aid is a long-standing topic of academic and practical relevance. A considerable portion of the economics (Mekasha and Tarp 2013; Sumner and Glennie 2015) and politico-economic literature (Bueno de Mesquita and Smith 2009; Neumayer 2003) has traditionally focused on the international distribution of aid. Recently, the better availability of geocoded information on ODA has allowed academics also to investigate the internal distribution of ODA within countries (Bluhm et al 2018; Briggs 2017).⁸

A strikingly consistent finding in this literature is that ODA flows to richer rather than poorer regions within a country. For instance, Öhler and Nunnenkamp (2014) have looked at the geographic distribution of World Bank and African Development Bank projects. For a set of 27 recipient countries they found that measured “needs”, such as high infant mortality, matter little in the distribution of project aid (Öhler and Nunnenkamp, 2014). Other studies concur (Briggs 2018).

Hence, within-country regional inequality is politically very salient, but project aid flows, it seems, do little to reduce inequality. This contrast motivates our research interest. How do aid agencies respond to regional divergence? Do they actively try to counterbalance regional polarization, or perhaps rather unknowingly reinforce it? To the best of our knowledge, scholarly research directly contributing to this is scant in number, but can make use of an extensive literature on “aid flows” between countries and domestic aid flows within countries.

First, bi- and multilateral donor initiatives have focused on poverty reduction in recent years. This should mean that aid flows to both the poorest people and the poorest regions. Yet for political reasons, the effectiveness and salience of poverty orientation in ODA has varied over the last decades (Bodenstein and Kemmerling 2015; Briggs 2017). For instance, a poverty orientation in ODA was very prominent in the 1950s and 1960s, but it only really came back in the early 2000s. In other decades, the focus on economic growth often prioritized economic hotspots while neglecting remoter, less productive regions.

There are well-known trade-offs in targeting the poor and poor regions (Besley and Kanbur 1990; Sumner 2012). Drawing on the literature about the determinants for the international distribution of aid, poverty and need are rarely the only or even the major concerns (Alesina and Dollar 2000; Bueno de Mesquita and Smith 2009). We do find similar results for the role of domestic or supranational forms of regional policy and aid (Kemmerling and Bodenstein 2006; Bodenstein and Kemmerling 2012; Arulampalam et al 2009). This literature discusses the role of internal inequality in the distribution of aid flows. For instance, De la Fuente and Vives (1995) examine the importance of equity vs efficiency criteria for the distribution of regional transfers and find that equity concerns matter relatively little. Even when donors emphasize the necessity of targeting poor regions and address the problem of regional inequality, the recipient’s motives also need to be aligned. Otherwise, donors face a dilemma: if they focus explicitly on regional imbalances of aid and try to target specific regions, the recipient might reject the intervention. Hence, there is a convoluted bargaining regarding what is acceptable for both sides, especially if the recipient is confronted with political problems of regional polarization. An extreme example is the problem of aid interventions in countries with ongoing violent conflicts. If donors press hard and explicitly against regional imbalances, this could increase the recipient’s resistance (Swedlund 2017) to let the donor channel money into “sensitive” regions.⁹

In this article, we show that the politics of targeting and the dilemmas associated with it are major reasons for when and how ODA will respond effectively to regional inequality. Donors have to balance the focus on poverty reduction and rural development with issues of practicality and the strategic interests that donors and recipients have. This prompts the empirical question of how donors react to regional inequality and difficulties in delivering aid to particularly poor domestic regions. Do donors perceive regional inequality to be a major concern? How do they even define or see regional inequality? And do they shift their aid strategy to specific topics (such as human rights, agriculture), specific sectors (basic assistance), or specific forms (project aid or humanitarian aid) to respond to regional inequality?

Myanmar as a case study of regional inequality

Myanmar is a particularly instructive case study to answer these questions because of its internal variation: some regions are harder to access than others, conflict areas shift over time, and there are long-standing politicized ethnic cleavages. Moreover, the drastic economic and – to some degree – political opening came suddenly in 2011 and ushered in a time-lapsed socio-economic transformation of the country (Asian Development Bank 2017; David and Holliday 2018; Thant Myint-U, 2019), while the political change itself proved to be superficial and short-lived (Campbell and Prasse-Freeman 2022). Nonetheless, such dynamics allowed for an enormous inflow of ODA from both Western and Asian donors (Carr 2018; Reiffel and Fox 2013). The abrupt and rapid pace of these changes allows us to investigate the mutual relationship of ODA and regional inequality more clearly than in countries where such processes happen much more gradually.

To understand how ODA relates to regional inequality today, we need to briefly consider the country’s historical context. Regional inequality and tensions between different regions run throughout Myanmar’s history. The agriculturally productive zones along the Ayeyarwady River have always been an attractive resource for local rulers and powerful neighbors (Harvey 2000; Scott 2009).¹⁰ British colonial rule and the immediate aftermath of World War II reinforced regional divisions and cemented the idea of two Burmas: a predominantly Bamar mainland under direct British control (“Ministerial Burma”) and the “Frontier Areas” endowed with greater autonomy (Charney 2009; Prasse-Freeman and Phyo Win Latt 2017; Thant Myint-U, 2001). Though intuitive and appealing in many respects views based on binaries (core versus borderlands) need some serious historical contextualization. Myanmar is home to several competing nation-building projects, but also several competing state-making ones, and there are multiple differences among the country’s many peripheries (Meehan and Sadan 2018). Crucially, there are also significant differences and hierarchies (racial, gender, class, ideological) in the core (Jordt, Than, and Lin 2021; Htet Min Lwin 2021). Actually, some of the divisions were and continue to be less regional than they are ethnic.¹¹ The Karens, the Mons and the inhabitants of Rakhine were not under the Frontier Areas, but were governed as part of Ministerial Burma. The highly centralized military governments after World War II governed the Bamar mainland and some remoter areas, while other areas such as Rakhine and Kachin states, among others, remained contested, home to alternative, non-state forms of governance. Even the decade from 2011 to 2021 did not bring the country any closer to peace. To a large degree, the political change exacerbated inequalities in wealth and the extraction of rents in an already highly unequal country (Campbell and Prasse-Freeman 2022; Calabrese and Cao, 2021). Thus, a nuanced understanding of Myanmar needs to refrain from excessively state-centric perspectives.

Some these binaries have tended to endure and have been even reinforced by the aid industry itself: a case in point is the dilemma in aid provision, because, by and large, donors need to pass through the gateway of the “dominant” nation-building project run by the internationally recognized sovereign government of Myanmar and its regime. In the Myanmar context, as examined by Décobert (2020 and 2021), Nilsen (2020), and Jung (2020), there has been a divide between development and humanitarian aid, with the latter going to community organizations in border and cross-border regions. State- and nation-building remain Myanmar’s long-standing and unresolved challenges. Such adversities also left their mark on the massive and increasing regional inequality.

Bearing all these caveats in mind, even brief experiences of traveling outside the largest cities reveals huge gaps in infrastructure and the provision of public services. Yet, even with all the cautioning against simplified dichotomies mentioned above, Figure 1 lends some support to this notion of “two Burmas” in which (economic) activity seems concentrated in the Bamar mainland, with other regions seeing hardly any of it.¹²

Figure 1. Change in nighttime luminosity between 1992 and 2012.

Notes: For data see text. Graph computed with QGiS (v.3.8.3) Software

The maps above show data on nighttime luminosity in Myanmar for two years, 1992 and 2012.¹³ The blue boundaries demarcate the states and divisions in the country. Even a cursory glance reveals three basic facts. First, Myanmar shows very low levels of nocturnal electricity consumption. This is true even when comparing its consumption levels to other poor countries in its neighborhood. Second, there are big differences between the Bamar mainland along the Ayeryarwady River (especially in the three main cities Yangon, Nay Pyi Taw and Mandalay) and the border regions. Third, these differences between the core and the periphery have increased over time: growing consumption mainly happens in the mainland, and specifically in the big cities.

While nighttime satellite data primarily show electricity consumption, it is also a valid proxy for measuring economic activity, especially for poor countries with unreliable administrative statistics (Henderson et al. 2012). Prospering economic regions, in turn, attract internal migration, and thus further exacerbate regional inequality. Based on these luminosity data, we can construct an index of regional inequality, measured as the standard deviation of average luminosity in each of the 15 regions. In Appendix A (Figure A2), we show some simple validity tests of the luminosity data with other sources of information. We also illustrate that Myanmar’s level of regional inequality is clearly severe but still not the extreme cases worldwide. This would suggest that the Myanmar case study holds interesting lessons also for other countries.

It is difficult to underestimate the growing salience of the regional differences in Myanmar, especially in terms of conflicts between a Bamar-dominated center and the horseshoe-shaped borderlands in which ethnic minorities predominantly live. As Jones (2014) aptly put it, in Myanmar “the periphery is central”. What follows is that regional inequality has three dimensions that overlap to a significant degree. First, the difference between capital and periphery. Historically the capital has shifted around between Yangon, Mandalay and Nay Pyi Taw, but there is a clear difference between these places and the rest of the country. Second, the difference between urban and rural areas: the Ayeyarwady River area including the three main cities is much more densely populated than the rest of the country. Third, the difference between Bamar and non-Bamar regions with all the ethnic conflicts this entails. It is clear that these distinctions are not totally clear-cut, but they are politically salient and hard to disentangle from one other. The question remains how donors, or even the recipient government, responds to this complex domestic regional inequality in the context of ODA flows.

ODA in Myanmar

Aid flows did not just (re-)start with the political reforms of 2011. Some donors never really aligned themselves with the US (and EU) sanctions imposed following the military takeover of 1988 and the crackdown after the 1990 elections. In addition, Cyclone Nargis in 2008 provoked considerable inflows of humanitarian aid over a short period. Nevertheless, it is clear that the opening after 2010 paved the way for a dramatic increase in the number of donors, the size of the aid flows, and the number of interventions (Reiffel and Fox 2013). The most drastic change came in 2012 to 2013 when, according to the OECD Creditor Reporting System, aid commitments to Myanmar increased from less than one to almost seven billion USD (Carr 2018, 5).

To assess the role of ODA in mitigating or accelerating regional inequality, we mainly rely on OECD/DAC data. We extracted information for the largest bilateral donors¹⁴ as well as the major multilateral ones.¹⁵ We also extracted information for Chinese aid flows from the website aiddata.org, but given its limited reliability and comparability, we used these data only as a memorandum item.¹⁶ We selected states and divisions and the capital as union territory as the level of geographic disaggregation mainly for pragmatic reasons of data availability, but also because the power balance between these units is a politically salient topic for the federalism of the country.

Putting all this information together yielded around 9,000 aid interventions between 1995 and 2015. We used the descriptions to code the geographic target destination in each intervention. Initially, we relied on human coding. In less than 15% of all cases, this yielded clear information about whether the intervention targeted one of the states and divisions or the central government. Therefore, we computer coded all those remaining interventions for which we did not have information with a battery of search strings. For instance, we coded rural projects where we found such terms in the project descriptions. We added to this category interventions that had the official Creditor Reporting System sectoral code for the primary sector (see Appendix A for details).

In total, we found reliable geographic information for almost 70% of all projects. As for the remaining cases for which no information was available, several possibilities remain. Either we could not allocate these projects according to any geographic criteria, or the criteria were not applicable (for example when money was dispersed to Burmese student scholarships abroad). A third possibility was that of multiple destinations of aid flows. We coded such projects with multiple destinations separately, but they resulted in a small fraction of the overall codings. Most of those projects for which we had no direct information arguably were destined for central government purposes. In this sense, our estimates of the regional share of ODA projects are an upper bound on the real (direct) regional share of all ODA projects and they (still) underestimate the share flowing to the central government.

To validate our results, we compared them with other sources. We used the official donor reports by the large donors and the Myanmar government for that purpose. Another important source for triangulation comes from a recent comprehensive survey of aid activities in Myanmar (Carr 2018). This detailed and meticulous survey computes the share of aid projects going to regions for one year only (2016), but the results for this year are very similar to results from our much simpler approach.

Figure 2 shows the dominance of the central government and capital city Nay Pyi Taw in receiving aid. Each bar represents the total sum of ODA inflow for the 20 years for which we had information. Out of a grand total of c.US$11 billion, the central government and the capital received about U$6.6 billion. Because Nay Pyi Taw is just the capital city of Myanmar, without much hinterland, it seems reasonable to lump ODA projects for the central government and for Nay Pyi Taw together. The region with the second largest aid inflows is Yangon, already below one billion USD. Rural development projects make up some US$1.4 billion.

Figure 2. Aid Flows (1995–2015) by Regions.

With these data, we have calculated three measures for the regional share of ODA for the 20 years between 1995 and 2015, focusing on the three dimensions of regional inequality mentioned earlier. Figure 3 shows the results for the regional shares. If we count only the share of all regions excluding the capital (“political center vs periphery”), the regional share spiked in the late 1990s, then fell and subsequently recovered to some 18% of all ODA in 2015. Adding all rural development projects (“urban vs rural”) which do not mention specific regions, we find similar movements ending in 2015 on a level of roughly 24% of all ODA. Finally, the share of money flowing to non-Bamar regions is much lower, hovering between 5 and 15% for most of the period.

Figure 3. Share of Aid Flows to Regions.

It is clear that inequality in Myanmar has several causes (Prasse-Freeman and Phyo Win Latt 2017, 409) and has many dimensions – from gender (Hedström and Olivius, 2022) to settlement types and ethnicity. To measure the impact of aid on inequality does not deliver results with our, rather coarse data, but it arguably falls behind other forms of capital inflows (Bissinger 2017). But, we are mainly interested in the opposite direction: It seems clear that donor agencies reacted to regional inequality. The combined share increased once regional inequality exploded. And yet we also see that, on an absolute level, the share that went to regions was still dwarfed by the sums allocated to the central government, and that there were differences between Bamar and non-Bamar regions. In the next section, we zoom in further on some of the donor interventions to examine in more detail how these agencies have responded to rising inequality.

We included all (major) types of aid flows in the analysis. This means that financial aid, especially in the form of debt relief is counted as aid to the central government. While this is not always accurate, the lion’s share of debt relief is channeled to the central government. In Appendix A, Table A2, we see that for some donors debt and emergency relief are, indeed, the largest forms of aid giving. Hence, the spikes account for large forms of debt forgiveness in 2012 and 2013, which made subsequent increases in aid possible, as detailed in Carr (2018). If we exclude debt relief from our analysis, the peaks are much smaller (they still happen because these years also showed big inflows of program and project aid), but the qualitative assessment of aid to regions vs. aid to the center remains.

How do donors respond to rising regional inequality?

In this section, we add qualitative evidence about whether and how aid agencies respond to regional inequality. To guide our reasoning, we explore three sets of questions. First, do donors talk about regional inequality, and, if so, how salient is it in the donor rhetoric? Second, does the money follow donor rhetoric (i.e. do flows also target more remote regions)? And, thirdly, what problems do donors confront when addressing regional imbalances? To answer these questions, we mainly used the donors’ official publications and complemented this information with (background) semi-structured interviews with representatives of donor agencies and Myanmar academics and experts.

We looked at different types of donors. Several authors have noted that aid giving in Asia differs in some ways from other regions. Stallings, (2017), for instance, have argued that Eastern Asian donors, such as Japan, are very different from their Western counterparts (see also Arase 2005; Carr 2018; Dreher et al 2011). For instance, Japan never really severed ties with Myanmar despite the US-led sanctions (Strefford 2016, 489) so they could quickly reopen aid channels once the sanctions were lifted. Also, Asian donors, such as South Korea and Singapore, are more interested in fostering regional economic exchange and a model of the developmental state compared to “Western” donors (Stallings, 2017, 16). Thus, we selected donors representing such different types: an old donor with a relatively new interest in the country (Germany); a new, rising Asian donor (South Korea); a donor with a long history with the country (Japan); and a donor with a long history of involvement interrupted by the sanctions (the UK).

An analysis of the donors’ country strategy papers revealed that very few among them refer to inequality explicitly, but most mention problems of specific, remote, and rural regions, as well as poverty in such regions. Germany is a case in point. Its major development agency GIZ (Gesellschaft für interationale Zusammenarbeit) returned to Myanmar after a 20-year break imposed by the sanctions. GIZ mainly implemented Germany’s development initiatives, backed by the local embassy and commissioned by the Ministry for Economic Cooperation and Development. GIZ’s priority areas were in the fields of education, sustainable economic development, and rural development.¹⁷ To help reduce poverty, GIZ primarily focused on developing the private sector, promoting the financial sector, and providing vocational training. The agency’s flagship projects were in aquaculture and electrification in rural areas. It is not always easy to break down aid data by a regional logic. When we look at our data, however, we do see that Germany’s share of aid going to regions was above average, hovering around 30% (see Appendix A, Table A1). We also see that German donors often targeted aid to NGOs in aid flowing to the regions, arguably trying to bypass the central government (Appendix A, Table A2).

While representatives of other donors noted the difficulty with disbursing funds and the low capacity of implementing projects, GIZ was more optimistic in this regard. GIZ deemed the local capacity in Myanmar to deal with aid relatively high.¹⁸ Neither did GIZ report any particular challenge when rolling aid to the periphery. In part, this could be the consequence of Germany’s aid not necessarily targeting borderlands per se and mainly focusing on relatively technical issues.¹⁹ The exception here was Shan state, where GIZ runs several aid interventions, such as investments in rural infrastructure.

Compared to Germany, the relatively new donor South Korea looked much less into the issue of regional inequality. South Korea’s development assistance to Myanmar was relatively small when compared to that of other East Asian countries but rapidly increasing (Fumagalli, 2017). The involvement of the Korea International Cooperation Agency (KOICA) in the country was wide-ranging, including support ($20 million) of the Myanmar Development Institute in Nay Pyi Taw to higher education projects with the University of Yangon as well as a $22 million-worth attempt to “export” Korea’s rural development experience through the New Village Movement (Saemaul Undong). Myanmar moved up in the priority list for aid recipient countries from being outside the top 20 in 2011 to number four in 2014.

KOICA cursorily mentioned the issue of inequalities among regions in its 2016 country strategy (KOICA 2017), but that was merely a short reference to the Myanmar government’s own economic policy document of 2016 (Government of Myanmar 2016). Meetings in both Seoul and Yangon²⁰ suggested that Korea was keener to follow other donors’ priorities than to promote an agenda of its own.²¹ In addition, the interviewees also emphasized that what matters to Seoul was to be aligned with the recipient’s interest. This was consistent with current donor practices that emphasized local ownership and coordination, and it effectively shifts the agenda setting to the recipient. A look at KOICA’s aid flows reflects this regional “indifference”, with most of the projects assigned to nationwide priorities (see Appendix A, Table A1).

Saemaul Undong was a rather short-lived exception to the rule. It drew on and referenced Korea’s rhetoric of “sharing” its own development experience, which has become a mantra of Korea’s foreign economic and aid policies (Fumagalli 2017; D’Costa 2015; Heo and Roehrig 2014). Some of the more generous contributions in the sector of rural development in Myanmar drew on former President Park Geun-hye’s own “obsession” with restoring the legacy of her father – dictator Park Chung-Hee – by promoting Korea’s own rural village development model across Southeast Asia. Its applicability and relevance to the Myanmar context were disputed both by Myanmar’s authorities and Korea’s own development agency officials,²² and were abandoned after the influence-peddling scandals of 2016—which incidentally also had a Myanmar dimension (Kim, 2018)—and Park’s impeachment in 2017.

The two long-standing donors, Japan and the UK, seemed to be more clearly concerned with the geographic distribution of ODA but in very different ways. Japan has been present in Myanmar for several decades, beginning with reparation payments after World War II and leading to massive inflows of Japanese aid (Reilly 2013; Steinberg 1991). The Japan International Cooperation Agency (JICA) ran projects throughout the country, from Kayah to Kachin state (Carr 2018). ²³ As a consequence, the share of aid targeting regions and the rural sector was higher for Japan than for all other bi- and multilateral donors in the country (see Appendix A, Table A1).

This high share going to regions seemed due to the high and uninterrupted levels of trust that Japan cultivated across all Myanmar authorities, which enabled JICA to operate in the borderlands. A review of the projects shows that – consistent with the broader aims and focus of Japan’s ODA policy (Arase 2005; Shimomura et al 2016)—the focus was on the promotion of local economic development through the support of infrastructural and energy projects. This focus seemed to make Japanese aid less politically sensitive for the Myanmar government. The regional focus, however, did not imply that money really benefitted the regions. The Baluchaung hydropower plants in Kayah state are a case in point. Kayah state is certainly in one of Myanmar’s poorest states; however, the projects employed very few locals, and the benefits for local communities and vulnerable people were unclear (Carr 2018, 12).

Compared to Japan, the UK – Myanmar’s second-largest bilateral donor – took a more explicit stance on the matter. The UK’s top priorities were comparable to those of other donors and included issues such as promoting good governance, inclusive growth, as well as peace building and national reconciliation. In addition, the UK also explicitly targeted “areas and groups affected by conflict which are deliberately disadvantaged” (DfID 2018, 2). The Department for International Development (DfID) further specified that this entailed “a geographical focus on the ethnic/border states” (DfID 2018, 2). Such an approach was unusual among larger donors, and it constituted a clear break with the British approach to Burma in colonial times. In practice, this meant that DfID closely monitored the situation in conflict-ridden Rakhine, Kachin, and Shan states.

The question, however, is whether DfID was successful in targeting poorer and conflict-ridden ethnic regions in Myanmar. Between 1995 and 2015, the majority of DfID money mainly benefited central destinations (see Appendix A, Table A1). Nonetheless, within the relatively small regional share, DfID did target mainly non-Bamar regions. For instance, some of the largest UK aid projects target emergency help in Rakhine state. Before the coup, the UK had 27 active projects with a specific or even exclusive focus on the borderlands. Only the “supporting peace-building in Burma” (identifier GB-1 -204,661, total budget just under £32.5 m)—the UK’s efforts to build the capacity of civil society, women, youth, and religious and ethnic communities through the Paung Sie Facility – was an exception, as the organization has worked across the country.

Of course, the overall low share to regions might be due to other reasons. For instance, the UK might have simply place more emphasis on the central government’s ownership or try to indirectly reach remoter regions by other means such as targeting sectors rather than regions. Nonetheless, the discrepancy between donor intentions and donor aid allocations suggests that donor rhetoric might sometimes create obstacles to targeting poor regions. Donors face issues in openly tackling regional inequality if they are confronted with a recipient for whom this interference is a political challenge. Donors using an indirect approach targeting specific sectors, such as Germany or Japan, might have more impact than donors that have an explicit approach. However, this comes with further costs such as relying on “non-political” aid and without control over where benefits of interventions eventual accrue. To illustrate this political dilemma further, we now switch to the perspective of the Myanmar government.

The recipient’s role: The case of reforming higher education in Myanmar

Given the politically sensitive nature of the topic, we could not directly interview government officials. What we instead use as empirical evidence comes from an aid intervention on a much smaller scale. We participated in a project of the Open Society Foundations’ (OSF) International Higher Education Support Programme (HESP). The project began in 2013 and ended in 2016.

The OSF project was the response, first, to the invitation of the U Thein Sein administration (Farrar (2013; IIE 2013) and, later on, the government led by the National League for Democracy (Government of Myanmar 2016). In this respect, the “opening of educational space” (IIE 2013, 5) was a fragile policy window for reform in the country (see Lall 2020: 16). The higher education system was derelict, underfunded, and poorly managed under the previous military regime (Esson and Wang 2018).

The curtailing in size and autonomy of higher education institutions was a deliberate reaction of the military regime to the student uprising of 1988 (Lall 2020, 132). The 1988 uprising led to the dismembering of the University of Yangon, with the opening of campuses in more remote parts of the city. The programmes most affected were all in the social sciences, an area the military government deemed to be breeding social unrest. It seemed a key priority for the National League of Democracy to reestablish the status of Yangon University as one of Southeast Asia’s gems of higher education.

The project’s focus was threefold: first to assist university and departmental leadership with curriculum reform and development and to work alongside individual staff on course development and issues pertaining to pedagogy through the introduction of new approaches to teaching and learning. This led to the reopening of undergraduate programmes in international relations at the University of Yangon after a hiatus of 25 years (since 1988) and later at the University of Mandalay, as well as the launch of new undergraduate programmes in political science at both universities.²⁴

Given that the highly centralized Myanmar government was mainly interested in creating national champions, the issue of regional inequality (between universities in Yangon/Mandalay and the rest of the country) featured neither in the project nor in the country’s general higher education reform. Stepping up the capacity of other universities, especially those in the borderlands, was neither a consideration for the government (Sadan 2014), nor for the donor as a consequence. In fact, the project was representative of broader international involvement in the country from 2011 onwards in that it was very much Yangon-centric, with the University of Yangon receiving most of the attention (Esson and Wang 2018; Phyu Thin Zaw 2017).

Spill-over effects only happened indirectly and unintendedly. One key feature of Myanmar’s highly centralized system in higher education helped the project’s “regional balance”. The central administration uses a policy of staff rotation, whereby faculty members are reassigned to different universities every four years or so. Such a turnover in staff allows the ministry to control universities to a large degree, but it also effectively means that potential investments in human capital will also be spread throughout the country. Due to this staff rotation policy, there were minor trickle-down effects of the OSF intervention to less-renowned institutions in Yangon, Mandalay, and beyond. Over time, and actually thanks more to word of mouth than a strategic approach on anyone’s side, faculty from other neighboring universities were invited to the workshops and began introducing some of the changes discussed during those events.

To summarize this rather small case, the Myanmar government did not have much interest in countrywide engagements that would actively seek the participation of universities beyond the Bamar heartland of Yangon and Mandalay (Sadan 2014). The ministry of higher education had other priorities, as evidenced by its statements, education reviews, and development plans for the higher education sectors. The University of Yangon, and to a somewhat lesser degree the University of Mandalay, were to be the priority. The donor, in turn, largely wanted to ensure the government’s ownership of the project. This was consistent with the emphasis in the aid industry on “local ownership” but also with the fear that otherwise the Myanmar government would have stopped the project.

Drawing on this brief qualitative exploration, we can now also see the donors’ dilemma in problematizing regional inequality. Against the background of a highly centralized recipient acting as gatekeeper, very few donors explicitly mention politically sensitive topics. Many do try to shift money into disadvantaged regions, either directly or indirectly. The exception here may be Korea, which maps most closely to the recipient’s interests. However, there are clear trade-offs. For instance, the Japanese (and perhaps German) route of focusing on relatively technical issues made it easier for them to reach remoter areas. This also often implies that the concrete benefits for the regions are unclear. In contrast, the UK exhibited a more explicit donor rhetoric, with a specific focus on non-Bamar regions, but then it faced problems in actually rolling out the money to the regions. A large part of the issue hence lies on the recipient’s side. Despite the political opening till 2021, Myanmar’s authorities still heavily guarded the regional redistribution of aid in the country.

Conclusion

Through a discussion of how both the national government and a select number of donors dealt with the question of regional inequality we contribute to the scholarly and policy discussions on the role of development assistance in addressing this issue. While regional inequality is an increasingly salient topic in many developing countries, international donors still need to pay more attention to it.

We used the case study of Myanmar to illustrate how the rapidly increasing regional disparity within the country puts the industry of development assistance under severe pressure. On the one hand, vulnerable target communities often live in the most peripheral regions; on the other, donor and recipient’s motives do not always align well enough to ensure an accurate targeting of this population.

An analysis of the geocoded aid interventions of major donor agencies in Myanmar over twenty years shows no strong trend toward poorer regions. There is some evidence that ODA donors have reacted to increased regional inequality in the most recent years, but aid flows still largely benefited central regions, especially if we look at all types of aid and not only project aid. Looking more closely at the discourse and practices of several donors, we identified some reasons why this is the case. Contrasting the UK experience with Germany and, in particular, Japan on the one hand and the South Korean experience on the other, we found that indirect ways of targeting (for example through specific sectors) can be more effective than explicitly targeting the periphery, where most of the ethnic minority groups are concentrated. Japan mainly relies on technical forms of aid and reaches more of the remoter areas, although it is not clear to whom the benefits really accrue. The experience of OSF’s involvement in Myanmar’s higher education reform confirms how difficult it is to address regional imbalances if the recipient’s motives do not align with the donor’s.

The military coup of February 2021 confronted donors with new dilemmas as to how best to continue to support the population without indirectly strengthening or even recognizing the military’s actions (Fumagalli 2022b). Just like in previous instances of military interventions, some countries and donors, controversially, sought to remain engaged, while others withdrew. Perhaps the pendulum will swing again from development to humanitarian aid, as it has before (Décobert 2020; Jung 2020).

To be clear, this study has a number of limitations. Geocoding aid interventions still confront researchers with several problems of geographic attribution and measurement errors. In this sense, our best guess is that the regional shares we produce are an upper boundary to the actual benefit for these regions. We acknowledge that our efforts are only a first and preliminary attempt and that with better data our picture will become more accurate. Second, we have relied on qualitative information, such as interviews and direct process observations. This revealed our own situated interpretation, and also meant that we depended on the responsiveness and transparency of donor agencies. Third, as discussed at the outset, we are fully aware that using preexisting categories such as “states & divisions”, “center vs. periphery” or “Bamar vs. non-Bamar”, no matter how commonly used, also in the aid sector, is problematic. A fully de-colonial approach will need to not just critique but replace those categories (Sadan 2020), so that the production of knowledge in the field of development assistance no longer relies on categories produced and reproduced in pre-colonial, colonial and post-colonial times. Furthermore, it is obvious that there are also many people in Bamar regions in dire need of support and that they are victims of the dominant classes in Myanmar themselves (Campbell and Prasse-Freeman 2022).

Given these caveats, the study’s main aim was to start a conversation about the relationship between ODA and regional inequality. We suggest that the Myanmar case holds valuable lessons for other developing countries in Asia and beyond. As we argued above, Myanmar is not necessarily an extreme case in galloping regional inequality and the political polarization this produces. Hence, it will be interesting to see whether and to what extent the findings from the Myanmar case carry over to other countries. Ensuring the “no harm principle” of international aid interventions is very difficult in such regionally polarized cases. From a policy perspective, we increasingly see that better regional targeting of aid within countries also shows beneficial impact on outcomes, such as lower infant mortality and perhaps even on emigration patterns (Kotsadam et al 2018b). Thus, there is still a lot to learn about the politics of targeting ODA effectively.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Arts and Humanities Research Council [AH/S00405X/1]; the Open Society Foundations; the Thyssen Foundation [Az. 50.14.0.046]; the Academy of Korean Studies [AKS-2015-R39]; and Universität Erfurt.

Notes

1. In this article, we use ODA and (international) aid interchangeably.

2. We use Burma to refer to the country before 1989 and Myanmar for the period after that, as this (formally, the Republic of the Union of Myanmar) is now the official name of the country. We use the term ‘Bamar’ to refer to the majority ethnic group. We use ‘Myanmar’ and ‘Burmese’ interchangeably as adjectives.

3. The transition was contested, its drivers manifold, its evolution uneven and unequal, in many ways reinforcing preexisting inequalities. While change was in many ways real, equally real were the polarization and inequalities that accompanied the process. It was a form of political liberalization and an instance of neoliberal capitalism for some in either the elites or dominant groups, and a multi-faceted form of structural violence for many others. What happened was an example of authoritarian resilience and non-pacted transition from above, with the final destination of that transition being less open-ended than some of the literature was ready to acknowledge.

4. If anything, the most recent Rohingya refugee crisis (Fumagalli 2022a) has exacerbated the situation. Parts of Myanmar have become among the least accessible territories for aid agencies (Carr 2018; Meehan 2015; Meehan and Sadan 2017).

5. We are aware that this is far from the only way for assistance to be provided. Humanitarian aid to Myanmar, for example, has focused on cross-border regions, particularly in the sectors of health and education (Décobert, 2020).

6. Administratively Myanmar is divided into 15 units: seven ethnically-defined ‘States’ (Chin, Kachin, Kayah, Kayin, Mon, Rahkine, and Shan), seven Regions (Ayeyarwady, Bago, Magway, Mandalay, Sagaing, Yangon, and Tanintharyi), and one union territory (the capital city, Nay Pyi Taw).

7. With reference to the Kachin experience Sadan (2013) complements the focus on the process of knowledge production in colonial times with a discussion of pre- and post-colonial periods.

8. There is some literature on ODA’s effects on household inequality (Castells-Quintana and Larru 2015; Herzer and Nunnenkamp 2012) but often with contradictory findings (Chong et al 2009).

9. Of course there are obvious administrative and logistical reasons why it is difficult to roll out aid to the periphery. Fragile states and polities with low state capacity are confronted with challenges in reaching all regions equally even if they wanted to. A pertinent example is provided by intra-state conflicts in which some regions are cut off from aid flows (Duffield 2012).

10. Clearly, there are also differences within the Bamar mainland, and the river delta, but these differences are still small compared to the ‘hinterland’.

11. We are grateful to one of the reviewers for focusing our attention on this aspect.

12. The very poor also live in the largely ethnically Bamar areas of the Dry zone and the Delta. We are grateful to one of the reviewers for pointing this out. Also, this is not to say that the borderlands do not experience any form of economic activity. Woods’s and Tharaphi Than' work on ‘ceasefire capitalism’ and borderland politics (Woods, 2011; Tharaphi Than, 2016), Sarma’s on the resource frontiers (Sarma et al. 2022), Aung (2018)'s and Prasse-Freeman’s research on the peripheries as sites of neoliberal capitalism and resource extraction (Mostafanehzad et al. 2022) offer important insights and multi-scalar perspectives on capitalism in the borderlands.

13. The data come from the meteorological service of the United States https://ngdc.noaa.gov/eog/index.html (for details see Appendix C).

14. Key bilateral donors include DAC countries such as Australia, Canada, France, Germany, Japan, South Korea, Norway, Sweden, the UK, and the US and non-DAC donors, such as Singapore, India, and Thailand.

15. Multilateral donors include the Asian Development Bank, the European Union, the Global Fund, the United Nations Development Programme, and the World Bank.

16. There are well-known definitional and operational questions concerning what China regards as aid that deserve separate analysis (see for example Dreher, Nunnenkamp and Thiele 2011; Stallings and Kim 2017).

17. As shown in https://www.giz.de/en/worldwide/11988.html.

18. Interview, GIZ, Yangon, 17 May, 2019.

19. Interview, GIZ, Yangon, 17 May, 2019.

20. Interviews in Seoul (January 2017) and Yangon (June 2018).

21. Interview with the senior representative of a Western donor agency, Yangon (May 2019).

22. Interviews in Yangon (June 2018) and Seoul (June 2018).

23. See also the map published by JICA (https://libportal.jica.go.jp/library/Data/PlanInOperation-e/SoutheastAsia/030_Myanmar-e.pdf).

24. OSF was not the only international agency involved in higher education reform. There were previous initiatives that helped to prepare the groundwork for launching the new programmes (see Lall 2020: 133; IIE 2013). Yet OSF’s presence was quite encompassing, and the organization quickly became the government’s main interlocutor in the higher education sector.

Appendix A

Geocoding aid projects in Myanmar

We used data from the OECD/DAC database (https://stats.oecd.org/Index.aspx?DataSetCode=crs1) and completed the data with data from Aiddata.org (https://www.aiddata.org/). We extracted information for all large donors (bi- and multi-laterals) over a period from 1995 to 2015. This gave us a total of more than 9,000 projects. We let a human coder code project initially, allocating code numbers for the 15 states and regions in Myanmar. We also gave a code to projects that clearly go to the central government, projects that mention multiple regions, those for which there was no information and those for which the human coder found too little information to decide. Next, for a small subset of the projects we asked a second coder to code the projects again to check inter-coder reliability. Intercoder reliability was high. Both the simple percent of agreement and Krippendorff’s alpha are higher than 80% (Freelon 2010). In the second stage, we used computer-assisted coding using word strings in the project titles and the sector code descriptions. For this purpose, we matched the data with the English translations of all the districts and townships in Myanmar (Myanmar Information Management Unit (http://themimu.info/baseline-datasets).) and ran search strings in the project titles to identify each time a district or township was mentioned. In the third stage, a human coder reviewed the codings again to check whether we misallocated some projects to the states and regions. As detailed in the text, we have missing values in those cases where our method did not produce any insights. These missing values are different from those cases in which we found that aid was destined for several regions or had an explicit but non-specific regional focus. As Figure 2 in the main text shows, this category is not very large.

Table A1. Shares for Different Donors.

Donor	Share of regions	Share of rural	Share of regions plus rural	Share of non-Bamar regions
France	0.10	0.01	0.11	0.03
Germany	0.30	0.01	0.31	0.19
Norway	0.07	0.01	0.08	0.05
Sweden	0.08	0.01	0.09	0.08
United Kingdom	0.12	0.00	0.13	0.11
Canada	0.14	0.00	0.14	0.13
United States	0.14	0.00	0.14	0.13
Japan	0.37	0.43	0.80	0.18
South Korea	0.07	0.05	0.12	0.01
Australia	0.12	0.02	0.14	0.09
IDA	0.08	0.00	0.08	0.00
ADB Special Fund	0.04	0.00	0.04	0.00
EU	0.09	0.00	0.09	0.09
UN Peacebuilding Fund	0.43	0.00	0.43	0.43
WHO	0.00	0.00	0.00	0.00
FAO	0.00	0.00	0.00	0.00
ILO	0.38	0.00	0.38	0.25
UNDP	0.15	0.08	0.23	0.14
UNICEF	0.02	0.00	0.02	0.02
UNHCR	0.00	0.00	0.00	0.00
UNAIDS	0.04	0.00	0.04	0.04
UNFPA	0.44	0.00	0.44	0.01
IFAD	0.00	0.00	0.00	0.00
Global Fund	0.00	0.00	0.00	0.00
Memorandum Item: China	0.33	0.00	0.33	0.17
Total Average	0.14	0.02	0.17	0.09

Table A2. Details About Selected Donors.

		Germany	UK	Japan	South Korea	Total
Most frequent	Aid Type	Project Aid	Project Aid	Project Aid	Technical Expertise	Project Aid
	Number (%)	135 (33%)	196 (52%)	1527 (59%)	566 (47%)	4671 (49%)
Largest Difference/Rural to Central Aid (%)	Aid Type (% to rural)	Aid to NGOs (75%)	Specific Purpose Programs (50%)	Project Aid (100%)	Specific Purpose Programs (80%)	Aid to NGOs (66%)
Biggest	Aid Type	Project Aid Debt Relief	Specific Purpose Programs	Project Aid	Project Aid	Project Aid
	Amount (mill.)	134	195	3117	461	7440
Biggest difference rural to
Most frequent	Sector	Education	Material Emergency Relief	Education Facilities	Primary Education	Multisector Aid
	Number (%)	74 (20%)	53(14%)	114 (4%)	158 (13%)	442 (5%)
Biggest	Sector	Transport Infrastructure	Material Emergency Relief	Electric Power	Road Infrastructure	Electric Power
	Amount (mil.)	20	140	596	135	1102

Notes: The table shows for each of the four selected donors as well as all donors several key indicators. For types of aid, it shows the most frequent category as well as the category with the biggest total amount. The table also shows where the difference between aid to regions and the center (including the capital) is the largest. For the sector it shows the most frequent sector as well as the sector with the largest total amount.

Appendix B

List of interviewees

Table

	Date	Position	Location
1	20 May 2019	GIZ Myanmar senior official	Yangon, Myanmar
2	18 June 2018	KOICA senior official	Yangon
3	1 November 2016	Taipei Economic and Cultural Office representative	Yangon
4	17 January 2017	Academic	Seoul, South Korea
5	18 January 2017	ASEAN-Korea Center	Seoul
6	20 January 2017	Asia-focused think tank	Seoul
7	26 June 2018	Businessman working in Myanmar	Seoul
8	6 August 2019	Former senior programme manager of large non-state donor	Online
9	8 March 2013	Yangon University, senior administrator	Yangon
10	15 May 2019	Academic staff, Dagon University	Yangon
11	Several 2013–2019	Mandalay University of Foreign Languages, senior administrator	Mandalay, Myanmar
12	Several 2013–2019	Mandalay University, senior academic staff, Dept of International Relations	Mandalay
13	Several 2013–2019	Mandalay University, senior academic staff, Dept of International Relations	Mandalay
14	12 May 2019	Peace-building NGO	Yangon
15	19 April 2019	Director, NGO focused on education in minority areas	Yangon
16	22 April 2019	Yangon University, senior administrator	Yangon
17	13 May 2019	Yangon University, senior academic staff	Yangon
18	13 May 2019	Yangon University, senior academic staff	Yangon
19	22 April 2019	Mawlamyine University, academic staff	Mawlamyine, Myanmar
20	22 April 2019	Inter-faith NGO	Mawlamyine
21	14 June 2018	Mandalay University, senior administrator	Mandalay
22	14 June 2018	Yadanabon University senior administrator	Mandalay
23	9 February 2020	Mandalay University faculty member	Mandalay
24	10 February 2020	Yadanabon University faculty members	Mandalay

Appendix C

Using satellite data

We use data on night-time luminosity from the US National Oceanic and Atmospheric Administration, which provides annual data for Myanmar regions and states going back to 1992. The original source is https://ngdc.noaa.gov/eog/dmsp/downloadV4composites.html. We accessed the data using AidData Geoquery tool. Using night-time luminosity clearly serves a proxy function. We tested the validity of this proxy by looking how much it correlates with other socio-economic indicators. We took the latter from the two UNDP household survey as well as preliminary results of the recent census (Government of Myanmar, 2014) as well as some ancillary sources (see Table A3 below). The cross-regional correlation is highest with indicators of poverty as well as access to electricity.

Table A3. The correlates of logged nighttime luminosity for 2012.

	1	2	3	4	5	6	Source
1 Luminosity (average)	1						see text
2 Poverty (incidence)	−0.73***	1					UNDP 2011: 15
3 Morbidity (incidence)	−0.63**	0.33	1				UNDP 2011: 77
4 Access to Sanitation	0.09	−0.15	−0.21	1			UNDP 2011: 66
5 Access to Electricity (in 2005)	0.053*	−0.41	−0.34	0.42	1		UNDP 2011: 67
6 Electrification (ADB)	0.61**	−0.32	−0.43	0.39	0.84***	1	ADB 2014: 145
7 Population Density	0.69***	−0.26	−0.44*	0.25	0.62**	0.77***	Government of Myanmar/UNFPA 2014: 10

Sources: 15 observations; average; UNDP 2011 = United Nations Development Program Integrated Household Living Conditions Survey (2009–2010); ADB 2014 = Asian Development Bank Myanmar. Unlocking the Potential. Country Diagnostic Study; Government of Myanmar/UNFPA 2014: Census Atlas Myanmar. The 2014 Myanmar Population and Housing Census.