Leapfrogging Manufacturing? Rwanda’s Attempt to Build a Services‑Led ‘Developmental State’

A key characteristic of East Asian ‘developmental states’ was a strategic focus on manufacturing, but developing countries today face more complex decisions about whether to prioritise manufacturing or attempt to ‘leapfrog’ straight to services. Rwanda is a striking example of a country that both resembles a developmental state and employs a service-led strategy. We argue that, unlike the Asian developmental states, Rwanda is not experiencing an integrated process of transformation that ties human development together with economic growth. Instead, there are two parallel trajectories within the services sector: a growth story linked to investment in ‘modern’ services (such as tourism, finance and real estate), and a human development story linked to ‘basic’ services such as health and education, with limited linkages between the two. This mismatch between growth sectors and the labour force raises questions about the sustainability of services-led approaches and their ability to emulate the East Asian experience. Résumé oppementaliste et qui utilise une stratégie axée sur les services. Cependant, nous sou-tenons que, contrairement aux États développementalistes asiatiques, le Rwanda ne connaît pas un processus intégré de transformation reliant le développement humain à la croissance économique. Au lieu de cela, il existe deux trajectoires parallèles dans le secteur des services: une croissance liée aux investissements dans des services «modernes» (tels que le tourisme, la finance et l’immobilier) et le développement humain lié à des services «de base» tels que la santé et l’éducation, avec des liens limi-tés entre les deux. Ce décalage entre les secteurs de croissance et la main-d’œuvre soulève des questions quant à la durabilité des approches axées sur les services et à leur capacité à s’inspirer de l’expérience de l’Asie de l’Est.


Introduction
Aspiring developmental states today are not only distinguished by their new geography but by varying approaches to structural transformation, and specifically the extent to which manufacturing lies at the heart of their developments strategies. The classical developmental state literature, which focused on Japan, South Korea and Taiwan, charted successful state-driven paths to structural transformation through manufacturing industry (Johnson 1982;Amsden 1989;Wade 1990). However, it is less clear to developmentally-oriented regimes in low-income countries today that manufacturing constitutes a desirable-or feasible-way ahead. The rise of Asian manufacturing, global value chains, global finance and new resource discoveries present a different range of possibilities and constraints.
In this context, ideas of services-led development have grown in significance. Service sectors have been prioritised in many developing countries for a variety of reasons, including deindustrialisation (Dasgupta and Singh 2007;Rodrik 2016), the hostile attitude of donor community to industrial policy (Rodrik 2008) and the diminished policy space resulting from World Trade Organization rules and other bilateral and multilateral trade treaties (Wade 2003). The idea of 'leapfrogging' manufacturing holds particular appeal in countries that are landlocked and face high import and export costs, making it harder to compete with existing low-wage manufacturing economies.
Despite this, evidence and opinion are divided on whether prioritising services over manufacturing is viable for sustained developmental progress. Some scholars have argued that developing countries' comparative advantages today 'can just as easily be in services as in manufacturing' (Ghani and O'Connell 2014, p. 3), and that a 'third industrial revolution led by services' (ibid, p. 2) is underway. Against this is the fact that historical experiences of successful developmental transformation without industrialisation are lacking (Chang 2002), outside of a few states heavily endowed with valuable natural resources. Moreover, recent research has shown that manufacturing still has unique 'growth engine' features (Szirmai and Verspagen 2015), and that manufacturing is as important as ever globally (Haraguchi et al. 2017).
In this paper, we contribute to these debates in two important ways. First, by offering a detailed case study of the experience of one country (Rwanda) charting a service-driven path to development. Existing research on the prospects for services-led development tends to focus on the cross-country level, seeking to identify general trends, despite the fact that the prospects for success in services relative to manufacturing are known to be highly context-specific (Eichengreen and Gupta 2011;Haraguchi et al. 2017). This must be rectified if we are to understand how service-led strategies relate to context. Second, we use the Rwanda case to explore the challenges arising from servicesbased development strategies and the degree to which they are able to achieve the kinds of economic and societal transformation associated with the manufacturingled developmental states of East Asia. While much academic analysis of the developmental state paradigm focuses on issues such as bureaucratic autonomy and the relationship between state and market, a relatively underexplored (but crucial) aspect of the East Asian experience has been the manner in which labour was integrated into development strategies. As we argue in this paper, services-based development strategies sit uncomfortably with this aspect of the paradigm, because of the type of skills and level of education required among the population to ensure that servicesled growth can be combined with broad-based, sustainable job creation.
Rwanda is an especially suitable case study in which to explore the relationship between the developmental state paradigm and a services-led approach. Both the structural conditions and aspects of the institutional context in Rwanda are reminiscent of the East Asian developmental states. However, the Rwandan government has chosen services rather than manufacturing as the key driver of transformation away from an agrarian economy. In this paper, we focus on two clusters of international services sectors, which, alongside tourism, have become important in Rwanda and other aspiring service-led developers: Meetings, Incentives, Conferences and Events (MICE), and Finance, Insurance and Real Estate (FIRE). Our analysis is based on qualitative interviews with a wide range of key informants conducted by both authors over a six-year period between 2011 and 2017. 1 During this period, service sectors have become increasingly visible elements of Rwanda's development strategy. This is especially evident through attempts to develop Kigali into a business hub through the construction of the Kigali Convention Centre and, more recently, through new tourism initiatives such as #VisitRwanda's $30 million sleeve-sponsorship investment in Arsenal F.C. Interest in manufacturing has recently been revived (Behuria 2017), but services contribute a larger share of growth and are growing at a much faster rate.
The existing literature on Rwanda's service-led approach is mainly econometric and uncritical in its approach to this strategy (Spray and Wolf 2016;Uwitonze and Heshmati 2016;Ggombe and Newfarmer 2017). In contrast, we argue that the emphasis on services, which has a clear rationale in the context of land-lockedness, donor pressures and limited natural resources, has contributed to economic growth but done little for equality and internal economic integration, and remains somewhat dislocated from Rwanda's labour market context. Rwanda has made significant investments in human capital (Lavers 2016;Williams 2017;Chemouni 2018), and made some accompanying gains in social indicators; yet we argue that, unlike in the classical developmental states, these processes are relatively disconnected from its economic growth trajectory.
Thus, the Rwandan narrative of growth and poverty reduction is underpinned by two distinct parallel trajectories, in contrast to the highly integrated trajectory of the East Asian developmental states. Part of this, we argue, relates to the focus on services because of the difficulty of generating the necessary skills and linkages to generate mass employment in the kinds of service sectors Rwanda prioritizes. This lack of integration has not been adequately acknowledged in research thus far, partly because it is also obscured by the way that 'services' sectors lump together extremely diverse forms of activity. A closer look at services, which recognises that economic growth in 'modern' services such as MICE and FIRE does not necessarily build on parallel developments in 'basic' services such as education and health, reveals the marked difference between Rwanda's trajectory and the East Asian model. Rather than the development of human capital forming a 'productivist' base for broader economic development, there is a mismatch between capital and labour skills, and a lack of attention to domestic linkages in the economy. Consequently, the structural transformation of Rwanda's economy rests on fragile foundations.

Services-Led Development and the Rise of MICE and FIRE
The Evolving Debate on 'Developmental States' and the Role of Services Johnson (1982) coined the term 'developmental state' in his study of Japan's Ministry of Trade and Industry and its strategic interventions to promote industrialization. The seminal works of Amsden (1989) and Wade (1990) on Korea and Taiwan followed. By the end of the century, a developmental state paradigm had emerged, though with significant differences in emphasis between key authors. 2 As mainstream development policy in the 1990s shifted towards policy alleviation, some developmental state theorists such as Evans (2014a) responded by beginning to incorporate human-centred approaches, in line with the ideas developed by Sen (1999). Other scholars were highly critical of this shift, which emphasised health and education services but did little to promote the production side of development or job creation (Amsden 2012). This split partially overlaps with a divide between arguments that continue to emphasise the centrality of manufacturing, and those that prioritise services, with Evans' (2014a) idea of 'capability-enhancing developmental states' in the latter category. In his recent contributions to the debate, he argues that 'the new centrality of services forces any state that wants to be "developmental" to focus more intensely on people and their skills, instead of machines and their owners' (Evans 2014b, p. 43).
This dispute obscures the extent to which the successful developmental states focused on the link between 'people and their skills' and 'machines and their owners': in other words, between labour and capital. Linking these two is especially hard in internationally-competitive service sectors as it is challenging to construct a comparative advantage in the requisite forms of labour. This raises questions about the extent to which a focus on such sectors can provide the 'potential for rapid technological progress and labour productivity growth' (Öniş 1991, p. 112), something that was central to the strategic selection of priority sectors in countries such as South Korea. In support of their drive to industrialise, the East Asian developmental states all employed varying forms of 'productivist' social policy, whereby labour was suppressed politically but the provision of education, healthcare and welfare were targeted towards the 'productive sectors' of society (Mkandawire 2007, p. 25). Social rights were thus initially extended only to industrial workers in South Korea (Holliday 2000), and 'aggressive, proactive manpower development strategies based on medium-to long-term vision of occupational skill requirements' were adopted in several cases (Bennell and Segerstrom 1998, p. 286). Meanwhile, in Japan and Korea, resources mobilised through pension funds were channeled directly into industrialisation, further driving a virtuous circle of social investment and industrialisation.
As a consequence of these strategies, the architecture of social policy was 'transformational', serving to integrate labour into specific industrial development strategies (Mkandawire 2007). These often-overlooked aspects of the East Asian experience were crucial for the rapid evolution of industrial productivity. They helped to generate high-yield labour that was 'able to reskill itself during the process of industrial upgrading' (Castells 1992, p. 53) and not only transformed the economy but 'completely changed the society' (ibid, p. 66).
The highly skilled and specialised nature of work in many service sectors poses challenges for creating the requisite forms and quality of labour, particularly for highly globalised international services sectors. Incentives set by external interests are partly to blame for this. In 1986, the World Bank argued that the social returns for higher education were lower compared to those of pre-primary education, an argument which laid the foundation for 'university bashing' from other donors and NGOs in Africa (Mkandawire 2011, p. 17). Donors' obsession with quantitative measures including enrolment and construction of classrooms to evaluate the effectiveness of their investments in education has also led to obscuring a focus on the quality of education, as Williams (2017) has shown in Rwanda. Thus, perhaps ironically, the attempt to create services-led developmental states faces particular challenges because of the greater difficulty in using 'basic' services (such as healthcare, education provision and forms of social protection) to generate an appropriate labour force for 'modern' services (such as finance, ICTs, tourism and real estate).

The State of Evidence on Services-Led Development
The classical pattern of structural change, theorised through the experience of industrialised countries, begins with the reduction of the share of output and employment in the agriculture sector and its transfer to the manufacturing sector, and later (in some advanced countries) to the services sector (Clark 1940;Chenery 1960). Though it could be argued that China has followed this model, India's decline in agriculture has been accompanied by some increase in industry but a much larger increase in services, which has been interpreted as a response to changes in the global economy (Nayyar 2013). These two paths of structural change have had varied impacts on the relationship between capital and employment in these countries.
Most African countries have seen a similar trend to India, involving substantial growth in the services sector even without the saturation of manufacturing demand. As of 2015, services contributed to almost half of Africa's total output and about one-third of formal employment (UNCTAD 2015). The increasing prominence of services has been accompanied by optimism about services-led growth. Much has been made of how information technology has permanently changed central characteristics of the world economy, with the rise of a network society (Castells 2010). Arguing against traditional Kaldorian (1967) perspectives, which favoured the central position of manufacturing in economic transformation, some now argue that services 'are leading to the expansion of manufacturing, rather than the other way around' Singh 2005, p. 1055). Clearly, some services-like finance-are essential components of economic transformation and growth. However, in today's more integrated world, and with the increasing prominence of global value chains, services that facilitate coordination, administration and transportation are considered strategically important (Kaplinsky and Morris 2016). 3 Ghani and O'Connell (2014) argue that not only do services comprise an increasing component of GDP as countries become richer but, in the contemporary global economy, they also create more jobs at earlier stages of development than manufacturing and demonstrate faster productivity growth. However, this positive assessment of services as a 'new boat that latecomers can take' relies on a broad and unspecified definition of services, with little consideration of either the informal economy or the nature of the jobs created. Crucially, the kinds of services that are driving growth and prioritised in policy are often very different from those in which jobs are being created. As has been argued elsewhere, cities in developing countries tend to exhibit a persistent disconnect between capital and labour (Schindler 2017), which is partly a function of low degrees of industrialisation.
Although services such as IT and finance can in theory play the 'escalator role that manufacturing has traditionally played', such services sectors are highly skillintensive and cannot absorb the type of labour that exists in low-and middle-income countries (Rodrik 2016, p. 29). Other services are either not very dynamic technologically or are non-tradable. How exactly 'basic' services such as education and health provision fit into ideas of service-led growth is also contested. These are usually seen as social goods that help build the human capital to spur growth in other sectors, but they also comprise significant growth sectors in some developing countries, particularly where international aid plays a substantial role in growing activity within these sectors. This underscores both how diverse the service 'sector' really is and the complexity of the stories of 'service sector growth'.
Moreover, there are important differences between 'first wave' service growth at relatively low-income levels, which tends to mostly comprise 'traditional' services (such as lodging, meal preparation, house cleaning, beauty and barber shops), and 'second wave' service growth, which includes 'modern' services such as financial, ICT, legal, advertising and business services (Eichengreen and Gupta 2011). Some developing countries have benefitted from 'second wave' service growth much more than others. Another report (UNECA 2015) argues that, since most services are nontradable, countries reliant on their services sector for economic growth will eventually struggle with trade balance constraints.

The Role of MICE and FIRE
Notwithstanding these ongoing debates, service sectors have become central to both growth trajectories and development strategy in many countries. Tourism in particular has become integral to economic strategies in many developing countries. International organisations like the United Nations World Tourism Organization (UNWTO) and academic scholarship (Wearing et al. 2005) have argued that the sector has potential for poverty alleviation. It also provides obvious benefits for developing countries in dire need of foreign exchange. However, a large portion of tourism scholarship has remained critical of the sector's developmental impacts (Harrison 2008). Despite the opportunities that tourism provides, small countries are invariably forced to develop strategies that leave them highly dependent on foreign capital and vulnerable to the perceptions of outsiders.
MICE has been an increasingly important component of growth of global tourism (Mistilis and Dwyer 1999). It refers to business tourism dedicated to planning and facilitating conferences, seminars, meetings, exhibitions or similar events. The strategic appeal of MICE is that it potentially links together multiple service sectors in a mutually supportive way, including tourism and hospitality, education, financial services, transport and logistics, and real estate development. MICE also boosts the construction sector, and is thought to have broader benefits in technology acquisition, knowledge exchange and the forging of business connections (Kumar et al. 2014).
MICE has become a component of economic strategies in several countries in the Asia-Pacific region (UNWTO 2012). Some developmental benefits have been recognised: for example, in the Philippines, MICE visitors spend an average of six times as much in the country as regular tourists (Francisco et al. 2017, p. 6). Although MICE provides foreign exchange and facilitates-associated infrastructure, prioritising it does little to accommodate low-skilled employment and, like tourism, success is highly dependent on foreign capital and perceptions. There is also evidence that international MICE tourists are unlikely to spend much time-or money-outside 'gateway' cities, limiting distributional benefits (Mistilis and Dwyer 1999).
A further group of services sectors can be categorised under the banner of FIRE. The study of the relationship between economic growth and the development of finance and insurance is beset with problems of reverse causality and contested data.
It has been argued that financial development is not only pro-growth but pro-poor (Demirgüç-Kunt and Levine 2009). However, one recent study of 21 countries in sub-Saharan Africa found that 'there is still very limited support for the hypothesis that financial development leads economic growth in SSA countries' (Menyah et al. 2014, p. 393). Significantly, financial sector growth has been preceded by the liberalisation of the financial sector in most African countries. In comparison, no East Asian developmental state relinquished control of their financial sectors prior to achieving significant structural change (Wade 1990). The casual relationships between insurance and economic development are equally contested, with one metaanalysis of 85 studies finding only very patchy evidence (Outreville 2013).
Real estate, the final component of FIRE, is a booming sector in many developing countries, often fuelled in part by aid, diaspora finance, high inflation rates and the lack of productive manufacturing sectors (Obeng-Odoom 2010; Goodfellow 2017a). This supports jobs in the construction industry and can encourage foreign investment. However, data from China suggest a much stronger causal relationship from GDP growth to real estate development rather than the other way around (Shen and Liu 2004). Evidence from OECD countries also suggests that FDI in real estate does little for economic growth (Gholipour et al. 2014). Growth aside, the developmental benefits of real estate development are not clear. While new models of state-led and market-driven forms of housing investment are unfolding across Africa, the former are often used for clientelistic purposes to reward supporters while the latter rarely provide for the poor, and most real estate development is untaxed (Croese 2017;Goodfellow 2017b;Pitcher 2017).
More broadly, growth in FIRE sectors does not generally translate into a proportional increase in employment. In the US, for example, the percentage of GDP generated in the FIRE sectors increased by 34.2% between 1979 and 2005, yet the percentage of the population working in those sectors increased by only 10.6% over the same period (Palley 2013). For over half a century, scholars have argued that this shift from the 'real' economy to the finance economy can contribute to economic vulnerability (Baran and Sweezy 1966). Overall, while there are obvious reasons why many developing countries prioritise MICE and FIRE today, these are mostly negative ones, i.e. their structural weaknesses and the difficulty of constructing comparative advantage in manufacturing, rather than clear evidence that MICE and FIRE will deliver developmental benefits.

Rwanda's Services-Led Development Strategy
The Rwandan Patriotic Front (RPF)-led government has overseen Rwanda's economic recovery since the genocide, despite widespread criticism of its record on human rights and political freedoms (Reyntjens 2013). Rwanda has enjoyed annual GDP growth of more than 6% most years since 1994. Although agriculture and industry have both experienced some growth, services sectors have been growing the fastest. The composition of Rwandan GDP shows signs of some transformation, with the main element of that being a shift from agriculture towards services ( Fig. 1). The share of services has increased from 29% in 1994 to over 47% in 2015, while industry has reduced from 21% in 1994 to about 14% in 2015. Figure 2 demonstrates that service sectors have had very healthy growth over the past 15 years, particularly trade and retail, finance, ICT and tourism. The primacy of such sectors was explicitly highlighted in the government's national development strategy-Rwanda Vision 2020, developed in 2000-demonstrating its eagerness to gamble on developing a knowledge-based economy (GoR 2000). Services were seen as becoming 'the most important engine of Rwanda's economy' (GoR 2000, p. 7). The developmental niche that the country aimed for was 'services, communication and financial sectors' (p. 16) to become a 'hub' at the 'Gateway between East and Central Africa' (p. 7).
In the more recent Economic Development and Poverty Reduction Strategy Paper II (EDPRS 2), there are some shifts in emphasis. While these include some attention to 'light manufacturing', this is listed alongside other 'emerging sectors' including business tourism, financial services and Business Process Outsourcing. Overall, the emphasis on 'modern' service sectors is still dominant, with the word 'manufacturing' appearing just 8 times in the 172-page document, relative to 33 for tourism and 33 for financial services/financial sector (MINECOFIN 2013).
Recent literature has emphasised the significance of services in Rwanda's recent development trajectory (Uwitonze and Heshmati 2016;Ggombe and Newfarmer 2017). This literature also highlights some key differences between Rwanda's experience and the more 'bottom-up' service-led growth in some major emerging economies. In India, for example, the advance of the service sector has been attributed primarily to the expansion of health and education in the population (Latha and Shanmugam 2014). In contrast, services growth in Rwanda has been driven by public investment (Ggombe and Newfarmer 2017) and factors such as increased access to finance, training personnel, new ICT applications and the tax system. In other words, it is a much more state-driven process (Uwitonze and Heshmati 2016).
Government officials explicitly state that prioritising services was as a result of their own domestic technical constraints (including transport costs and limited domestic market). 4 Transforming Kigali into an international 'hub' of various kinds-transport, financial, tourism and information technology-was seen as integral to linking Rwanda to the fastest-growing sectors of the global economy. Tourism has become an increasingly important source of foreign exchange; by 2015, it had become the single largest export sector, accounting for 29% of all exports (IGC 2016). In 2014, the government received US$342 million from tourism receipts (Bishumba 2016).
There can be little doubt that the Rwandan government has worked particularly hard to grow its tourism industry. They invested in the construction of what is now the Serena Hotel (against donor advice), hoping it would work as a 'demonstration effect' to bring international hotels to the country (Behuria 2015). The government has also heavily invested in the national airline (Rwandair), the Kigali Convention Centre and in conservation attempts such as the Virunga National Park (VNP) and other tourist destinations within the country. The Rwandan government took the decision to prioritise the tourism sector as part of following advice from donors and consultancy groups, with a National Tourism Strategy developed in 2001 (Mazimhaka 2007). Supporting gorilla tourism-historically, Rwanda's most significant tourist attraction-was central to the strategy. The dependence of tourism revenues on gorilla tourism increased from 75% in the early 1990s to 93% (Mazimhaka 2007). Of the 800 mountain gorillas residing in the world, 440 are in Rwanda, with 216 born since 2005 (Tashobya 2016). The government acknowledges that gorilla tourism 'is our main attraction', 5 and, in line with leveraging growing international interest, gradually increased the cost of a permit for foreign visitors from US$250 in the early 2000s to US$750 in 2013. Over 20,000 people visited the VNP in 2014, more than three times the number that visited in 2003. Building on the apparent willingness of foreign visitors to pay high prices, in 2017, the government doubled the cost of a permit from US$750 to US$1500. Operators and consultants argued that this insistence on high prices relative to neighbouring Uganda and DRC was part of 'an obsession with high-end tourists', which neglected the other kinds of opportunities available. 6 However, the 2017 decision also demonstrates efforts to build Rwanda into a broader tourism and MICE destination; thus, a 30% discount applies for visitors who visit other national parks in Rwanda for at least 3 days, and a 15% discount for those who stay 'pre-and post-MICE events' to see the gorillas. 7 (Fig. 3).
This reflects a broader aim became to 'go beyond gorillas' and establish Rwanda as more than an add-on destination at the end of a tour of East Africa (GoR 2009). Emphasis was placed on attracting 'high-end international tourists' and, by 2010, RDB officials were impressed with the increases of visitor numbers in national parks and began to see MICE as a priority area for further support, 8 including through improving customer service in hotels and restaurants and expanding air travel facilities. 9 Currently, the average visitor to Rwanda spends only 3.2 days in the country, and it is seen as crucial to try and raise this to a week or more (IGC 2016, p. 25). In pursuit of this, Rwanda sharpened its focus on the key sectors needed for its MICE ambitions to succeed. In the following section, we examine how the government has built the MICE sector, and how this has been supported by a concurrent focus on FIRE.

Chasing MICE
The Rwanda Convention Bureau (RCB) was launched by the government in 2014. Rick Taylor, CEO of the Business Tourism Company that helped to set up the RCB, argued that the 'propeller' to deliver the Rwandan government's vision of becoming the first middle-income country in Africa 'is categorically, the MICE sector.' 10 The government claimed success in the few years after RCB was created, announcing that MICE revenues were projected to triple from US$49 million in 2013 to $150 million by 2017. 11 In the same period, the government was also working feverishly to complete the physical centrepiece and symbol of its MICE strategy, the Kigali Convention Centre (KCC). The project was initiated in 2007 by three corporate investors-Rwanda Investment Group, Prime Holdings and the Rwanda Social Security Board (RSSB). Initially slated for completion in 2011, it was beleaguered by controversy, escalating costs and terminated construction contracts, and only finally opened in June 2016. Expectations for the Convention Centre were piled high: a government official argued that it was a 'gamechanger for Rwanda's development', while a private consultant characterised it as 'Rwanda's version of the Eiffel Tower. That's how important it is.' 12 With construction costs of over $300 million, the KCC is now the most expensive building complex in Africa. The KCC comprises a 5-star Radisson Blu hotel, a 2600-capacity conference centre, the Kigali Information Technology Park and a museum on the bottom floor.
The effort and expenditure to facilitate MICE activities have been reflected in increasing numbers of high-profile events. In 2016, Rwanda hosted over 40 international conferences, including high-profile events such as the World Economic Forum, the Global African Investment Summit and the AU Summit. Consequently, Rwanda was ranked joint third in the International Congress and Convention Association (ICCA) 2016 business tourism rankings for the continent, after South Africa and Morocco. This actually exceeds the RCB's own ambitions, which were to rise into the top 10 by 2016, up from 21st in 2014. Kigali itself is now placed third among the continent's top cities for meetings, conventions and events, above Durban and Johannesburg. In 2016, Kigali hosted the same number of major meetings as Mumbai, Basel, Philadelphia and Nanjing (ICCA 2016).
By the end of 2016, Kigali had 3400 upper-and middle-range hotel rooms available. 13 Hotels had been another major focus of the government since 2007, both in the capital and beyond. 14 One businessman recounted how the President invested his own money in regional investment groups to persuade local businessmen to invest in hotels in regional towns. 15 In Kigali, a prominent local businessman, who previously served in the Rwandan Patriotic Army (RPA) during the liberation effort, led investment in the first Marriott hotel in sub-Saharan Africa outside of Cape Town. This was a further landmark for the MICE sector, as it contains 18 conference rooms. Alongside the various major investments in hotels, Rwandair, the country's national airline, significantly expanded its fleet, providing infrastructural support to the MICE industry. The government owns 99% of Rwandair, which began operations in 2002 and now operates routes to over 20 cities within and outside Africa. Rwandair is yet to make a profit, but government officials argue that it is integral to promoting Kigali as a services hub in Africa. In addition, the Rwanda Association for Professional Conference Organizers (RAPCO) was established to develop an accreditation system for professional conference organisers. This, it is hoped, will build on the positive reputation Rwanda enjoys with large shares of the global business community, evidenced by Rwanda's 2015 ranking in the World Bank's Doing Business Index as the third-best country to conduct business in Africa, as well as third-least corrupt country in Africa by the Global Corruption Perception Index (Rwirahira 2017). Criticisms of the Rwandan government from within academia (Reyntjens 2013;Thomson 2013) have not damaged the country's reputation among the global business community, providing a further boon to the MICE strategy.

Sparking the FIRE
FIRE sectors are natural counterparts to MICE, particularly given the major, capitalintensive infrastructure and real estate requirements for MICE activities. 17 Financial liberalisation rapidly gathered pace in the 2000s, as a number of foreign banks bought shares in struggling Rwandan commercial banks and new commercial banks from Nigeria, Kenya and Uganda entered the market . In 2017, there were 13 commercial banks in the sector. The government only retains a majority share in the largest bank, the Bank of Kigali (BK), which enjoys 30% of the financial sector in most indices including assets, deposits and liabilities. Despite BK's influence, commercial bank representatives themselves commented on their surprise that the government eagerly liberalised finance before 'Rwanda can stand on its own two feet.' 18 As the liberalised financial sector expands to include  17 We do not discuss the insurance sector in Rwanda in detail, both for reasons of space and because it is considered to be performing relatively poorly and lagging behind other service sectors (http://www.newti mes.co.rw/secti on/read/20840 2/, accessed 18 July 2017). 18 Interview, foreign bank official, January 2015. more citizens, the loan books of commercial banks have developed a concentration that reflects an emphasis on MICE and FIRE. In 2016, construction, hotels, commerce and transport accounted for more than 85% of BK's loan book. Commercial bankers' loan books had a similar composition, and some were worried about this concentration, highlighting that it was 'a ticking time bomb' and 'unsustainable in the long run.' 19 Simultaneously, the government has put significant effort into developing the real estate sector, with a strong focus on Kigali, and-as with the tourism sector-a strong focus on the 'high end'. This includes, most spectacularly, the 'Vision City' project, the first phase of which comprises 504 variable-sized units that retail at prices ranging from US$370,000 to $560,000. 20 When complete, the project aims to accommodate over 22,000 people, and is reputedly the largest real estate project in East Africa. 21 This project is funded by the RSSB, which has been pivotal in the expansion of MICE and FIRE, since it holds 40-50% of liquidity in the financial sector and its investments are largely concentrated in real estate, hotels and construction. 22 Also investing in real estate are domestic party-owned enterprises such as CVL Developers (the real estate arm of party-owned conglomerate, Crystal Ventures), Kenyan firms such as Fusion Capital and some domestic private investors.
There has also been a rapid increase in mortgage provision in recent years. Up to 70% of the loans made by some banks are now in the form of mortgages. 23 There is concern about the sustainability of this model in the Rwandan context; some bankers have allegedly 'grown wary of bricks and mortar.' 24 Interest rates of around 17% combine with the costs of landlockedness, and the removal of construction incentives in 2013, to render construction very expensive in Rwanda. A 2017 report from the consultancy firm Turner & Townsend (2017) found construction in Kigali to be more expensive per square metre than in Johannesburg, Kampala, Dar es Salaam and Nairobi. At $1117 per square metre, the cost is approaching twice that of Nairobi's $683 (ibid). This makes building for the mass market very difficult, and a handful of real estate developers in Kigali are all 'fighting for 20% of the market', with very little provision for the majority. There is increasing evidence of oversupply at the high end, with rents starting to dip significantly from around 2014, despite a chronic housing shortage. The significant growth in construction and real estate thus conceals growing concerns about a bubble in the process of bursting (Goodfellow 2017a).
These recent developments reflect the fragility of Rwanda's services-led development strategy. The success of MICE and FIRE depend significantly on 'going big'; in other words, capturing high-end, luxury international markets. Even beyond 19 Interviews, commercial banks, June 2017. 20 Documents obtained from Ultimate Developers, June 2014. 21 Interview with foreign construction firm A, January 2015. 22 Interviews, commercial banks and government representative, August 2016 and June 2017. 23 Interview with banking official, January 2015. 24 http://www.afric aprop ertyn ews.com/east-afric a/3249-prope rty-bubbl e-looms -in-rwand a.html, accessed 18 July 2017. the sectoral focus itself, Rwanda's approach differs significantly from the classical developmental state approach in ways that are intrinsically linked to its decision to focus on international services. These include the liberalisation of finance and the outward-facing role of the RDB, which acts primarily as a hub for investors to polish Rwanda's international image rather than a nodal agency stimulating innovation and participating directly in industry (Behuria 2018). As we explore below, this services-driven strategy also involves significant differences in the capital-labour relationship, with implications for the potential for transformation of the kind achieved by Asian developmental states.

Parallel Development Trajectories and Missing Linkages
While they undeniably played a role in economic growth, the contribution of service sectors such as MICE and FIRE to employment and internal economic integration have been much more ambiguous. The 'black box' of services conceals this to some extent: because services have increased substantially as a share of both GDP and employment, it looks as though a virtuous circle is at play. However, a closer analysis reveals gaps and vulnerabilities, raising questions about the sustainability of the approach. It is helpful here to distinguish broadly between three types of service sectors (building on the existing literature): 'basic services', such as health, education, and forms of social protection; 'traditional services', such as small-scale provision of low value-added services in the informal economy; and 'modern' services, such as those discussed above. Rwanda has seen substantial investment-fuelled economic growth in the latter, alongside largely aid-funded progress in the first category (Chemouni 2018; Williams 2017). It has also experienced what we might characterise as ordinary, expected growth in the middle category, which tends to swell in the context of urbanisation. However, the lack of an integrated development trajectory linking the three into a process of deeper structural transformation is clear from a closer qualitative analysis of what lies behind the statistics.
For example, it is still 'traditional' service activities such as wholesale and retail trade, repair of motor vehicles and motorcycles, and transportation and storage that comprise the majority of establishments and employment in the services sector (NISR 2017). Similar to many other developing countries, these are the kinds of activities that swell the numbers of people working in services. They are also sectors in which the vast majority of workers, 93-98%, according to the government's own data (NISR 2017, p. 136), are considered to be working informally, with little connection to the MICE and FIRE strategy. There is therefore a gulf between service sectors that employ large numbers of people and those that are leading economic growth. When it comes to contribution to GDP, the MICE and FIRE sectors loom large: real estate, for example, comprised almost 8% of GDP in 2017, more than the entire manufacturing sector at 6%, but employs a mere 0.1% of the population (BNR 2018;NISR 2017, p. 136). Finance comprised around 3% of GDP in 2017 (four times the amount in 1999), yet finance and insurance combined employ only 0.4% of the population (ibid) and just 1.75% of the people working in 'services' (BNR 2018).
On one level, it is not surprising that the 'modern' services sectors employ relatively few people, and the increase in output from these sectors relative to employment is often celebrated as 'increases in productivity' driven by a sectoral shift out of agriculture and towards services (e.g. Ggombe and Newfarmer 2017). However, the fact that these are capital-rather than labour-intensive sectors can be seen as highly problematic in the context of Rwanda's demographic shift, with the working age population projected to increase from 6.7 million in 2014 to 16 million in 2050 (World Bank 2016), and contrasts starkly with the 'developmental state' experience. In finance and insurance, for example, the fact that 75% of people are formally employed and 53% have a university degree only underlines the gulf between these people and the vast majority of those working in 'services', a gulf that Rwanda is finding extremely difficult to bridge.
Tourism provides an interesting example. In 2016, it contributed approximately 74,000 jobs, adding 123,000 jobs indirectly. 25 However, despite the fact that it has been the most successful in creating jobs of all the 'modern' service sectors, the sector retains an 'acute gap of 69.4% skilled human resources' (MasterCard Foundation 2016). Attempts to provide skills training for the tourism sector have lagged behind the needs of the sector. The government established Rwanda Tourism University College (RTUC) in 2006, and the Akilah Institute for Women in 2010, but such initiatives are very small in scale and have not provided the requisite skills at the pace needed to match the growth of tourism investments. Foreigners dominate senior and managerial roles in the sector, with one manager of a five-star hotel in Kigali noting that 'tourism gives some employment but we also have to compete for jobs with Kenyans and Ugandans. Most of the managers are also Kenyan.' 26 The skills deficit is especially acute in the hotel and hospitality sub-sector, as well as in sub-sectors such as passenger air transport (RDB 2012a). Additionally, 50% of staff were thought to need language training in key international languages including English, French and Chinese (RDB 2012a, p. 70). Other basic skills were seen as lacking: in the words of one tourism operator, 'Graduates from RTUC know how to fold a napkin but they don't actually know anything about tourism or service.' 27 In general, there is an emphasis on certification rather than developing the problemsolving, project management, entrepreneurial and team management skills needed by managerial staff (MasterCard Foundation 2016).
Similar problems can be observed in other MICE and FIRE sectors. In financial services, a bank official noted that 'our biggest problem has been human resources', due to a lack of skilled financial experts who can be hired 'at the right price.' The tendency therefore is to hire foreigners, usually from elsewhere in the East Africa region because 'it is hard to hire Rwandese.' 28 Meanwhile, there are fears of a brain drain because Rwandans returning home with international Masters degrees are 25 Internal consultancy report. 26 Interview, hotel manager, March 2012. 27 Tour operator quoted in MasterCard Foundation (2016, p. 16). 28 Interview with banking official, January 2015. often unable to find suitable work, and end up in 'lowly jobs' such as cleaning. 29 Paradoxically, qualified Rwandans therefore often find themselves applying for jobs elsewhere, while firms resort to hiring foreign staff to get the specific expertise they need. This disjuncture suggests a mismatch between the skills being acquired by educated Rwandese and the sectoral demands generated by new investment.
Construction firms echoed the view that it is 'a huge drain on resources to hire locals.' 30 The skills shortage in this sector, which is usually considered within 'industry' but underpins both MICE and FIRE, is particularly acute, amounting to 65% of the existing labor force. Despite increasing demand for professional construction services, 'the number of students taking construction courses and graduating from universities is dismal. In the last two years, less than 130 professionals have graduated in this field.' (RDB 2012b, p. 78). This represents a clear missed opportunity to link labour to the booming sectors of the economy. Moreover, in this sector especially, there is controversy around employment due to the role of Chinese investors and contractors. A common perception was that Chinese firms 'bring in hundreds and sometimes thousands of Chinese workers', partly because 'the construction manuals are all in Chinese', 31 and that 'the skills transfer just isn't happening yet.' 32 Chinese construction firms contest these narratives, arguing that one Chinese supervisor will oversee as many as 30 local staff, and that they are committed to building local capacity. 33 The Rwandan government itself has always taken the attitude that 'we can buy capacity while building our own.' 34 However, in the relatively liberalised MICE and FIRE sectors, the extent to which 'buying capacity' and building it are mutually compatible is questionable: without adequate training or skill transfer, the buying-in of capacity could undermine building it at home.
Some systemic impacts of MICE on internal economic integration may also be undermining its growth benefits. Despite claims that job creation in major international hotels like the Serena generates multiple additional jobs in he wider society (MasterCard Foundation 2016), there are significant drawbacks to the proliferation of high-end hotels. Some are now over-supplied and damage mid-tier hotels that were 'already operating at or below 50% capacity utilisation […] below the breakeven point for many ' (IGC 2016, p. 25). Between 2013 and 2015, at least 100 hotels were put up for sale after owners failed to repay their loans (Gahigi 2015). The government's hotel strategy does not currently match the demand of visitors, aside from the brief duration of occasional international conferences. Even long-standing hotels such as Mille Collines were only around 30% full most of the time in 2015. 35 Other concerns relate to the extent to which robust backward linkages are being developed in the construction of hotels and other MICE and FIRE projects. MICE strategy is characterised by a reliance on foreign investors for large-scale investment, alongside some strategic use of state-or party-owned enterprises and a handful of domestic entrepreneurs. There is some evidence that these firms enter partnerships with smaller local firms, for example in terms of inputs into the construction sector. 36 However, in the rush to feed the construction boom, many Rwandese actors claim that the government eagerly hands contracts over to foreign firms without sufficient attention to integrating the domestic economy. In the words of the CEO of one Rwandan firm: Rwanda is the only country I know where you can just sit on your computer in China and bid for something and win it…even in Kenya, you would have to have local content, prove that what you were planning to do can't be done by locals…we need to start shifting in this direction. 37 Many large foreign firms are not registering with the relevant professional body in Rwanda-the Private Sector Federation (PSF)-in a way that enables linkages, knowledge exchange and skills transfer to occur. There is therefore no mechanism to ensure large foreign firms share their learning and skills. This was corroborated by interviews with foreign firms: one representative of a Chinese firm was not even aware of the PSF and had never been approached to join it. 38 This raises broader issues about linkages in the MICE and FIRE sectors and whether these are being cultivated, including between MICE, manufacturing and agriculture, which have greater unskilled job creation potential. In 2008, OTF argued that Rwanda must develop 'stronger bonds between large hotels or "anchor firms" and local SME providers, such that goods and staff are sourced locally.' 39 However, some small firms suggested the government was too focused on identifying strategic priority sectors and did not help entrepreneurs develop linkages, for example by encouraging the manufacture of goods to feed into MICE sectors. 40 Given these problems, it is questionable whether Rwanda's bold attempt to push a services-led strategy through a developmental state approach can generate the kinds of structural and societal transformation associated with the East Asian experience. The combination of low wages with rapid increases in productivity was crucial in East Asia and was supported by productivist social policies that channeled welfare and human development directly towards economic goals. This is harder to achieve with services like MICE and FIRE. People with the requisite skills are few and demand high wages, with international staff filling many of the gaps. Unlike in Taiwan, where the proliferation of highly localised small industries provided opportunities for low-skilled and low-wage factory work that could then be progressively up-skilled and up-scaled as the economy developed, Rwanda is attempting to leap into high-end, high-skilled activities. 36 Interview with construction firm Project Manager, January 2015. 37 Interview with Director of a domestic firm, January 2015. 38 Interview with Chinese firm representative, January 2015. 39 Internal consultancy report. 40 Interview with entrepreneur, January 2015.
None of this is to say that Rwanda has not invested substantially in human capital formation: it clearly has. It can even be argued that its approach to social protection programme has 'productivist' elements, in the sense that it goes beyond basic poverty-reduction goals to support a broader development agenda, as Lavers (2016) has shown. However, the function of social protection beyond poverty alleviation in Rwanda is more about promoting community self-reliance and securing government legitimacy than creating a productive workforce for specific sectors (ibid). Meanwhile, advances in healthcare, such as the community-based health insurance, remain heavily donor-dependent, pursuing an elusive goal of universal basic coverage (Chemouni 2018) with little link to a productivist agenda. Even clearer in support of our argument is the case of education. Williams (2017) has shown how performance-based incentives led to progress in education in terms of quantitative measures like construction of classrooms and enrolment, but the quality of education and its capacity to produce a skilled workforce was very poor. Thus, school-leavers find themselves 'included in the country's development project vis-à-vis the education system while simultaneously excluded from meaningful opportunity given the poor quality of that system' (ibid: 559). The trajectories of human development and economic growth in Rwanda are therefore largely progressing along separate, parallel tracks rather than being progressively intertwined and mutually supporting each other, as was the case in East Asia.

Conclusion: Prospects for a Services-Led Developmental State
Rwanda is attracting attention as an aspiring services-led developmental state with a rapidly growing tourism sector, and success in the MICE and FIRE sectors. Although the government has directed this services-driven approach through state-led strategies reminiscent of developmental states and has achieved substantial growth, its priority has been to develop links outwards with global finance and engage with the international business community. Less attention has been paid to the needs of businesses, the role of labour and the potential for creating domestic economic linkages.
The Rwandan experience raises important questions about how a focus on highend services is facilitating internal integration, technology acquisition and skills transfer. These issues are often overlooked in the literature on services-led versus manufacturing-led development, which is largely concerned with short-term growth effects. The focus on 'going big' across a range of interlinked services sectors in Rwanda has a certain logic. But one of the key difficulties in pursuing this approach is that international service sectors have increasingly globalised standards and processes, which Rwandan training institutes are not yet seen to be providing. Time is also a significant constraint, given that the rapid growth of MICE and FIRE sectors have left little scope for local Rwandans to acquire the skills necessary for quality employment in those sectors. Thus, even where there is substantial job creation resulting from investments in these sectors, many jobs are not filled by Rwandans.
These problems are likely to be faced by any country seeking to specialise in 'modern' international services from a position of low levels of human development, and are particularly worrying for those with large and growing youth populations. The dynamism and highly globalised nature of these sectors means that trying to catch the 'boat' requires rapid, focused and targeted investment, not only in terms of capital but also in labour. This is much harder to do than in the context of manufacturing industry, which creates more low-skilled than high-skilled jobs. What is likely to happen in the case of 'modern' services is the creation of relatively few highskilled jobs and even fewer low-skilled ones. Parallel efforts to provide education and social protection for the majority cannot easily bridge the gap either by creating lower-skilled jobs or by generating the requisite skills to fill the few high-skilled ones.
Rather than producing an East Asian-style outcome, these parallel trajectories are likely to produce volatile growth coupled with significant inequality and a reliance on that other group of 'services': traditional precarious labour associated with work in the informal economy. A services-driven development route necessarily involves a fundamentally different trajectory from the classical developmental states. Servicesdriven development can potentially succeed, but there is no automatic 'trickle-down' from 'modern' service growth. Any broader development depends on continuous redistributive policies, which will probably need heavy support through continued aid. The Rwandan government has belatedly realised that a better balance between services and manufacturing is required. However, any country hoping to emulate the 'developmental states' has to work hard to innovate in building the linkages between manufacturing and services, and between both of these and its labour force.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.