Legacies of state-building and political fragility in conflict-ridden Yemen: Understanding civil service change and contemporary challenges

Abstract Constitutional, economic and administrative reforms have generated debates about their unintended consequences in poor and developing countries, and the best way to steer them towards better outcomes. In-depth case study analysis helps in tailoring future reforms and enriches academic literature on countries faced with the complex and intertwined problems of fragility, traditional actors, state building and donor involvement. This paper aims at examining the legacies of state-building and civil service reforms in Yemen by providing a narrative of motivations, outcomes as well as the involved politics and actors. Yemen’s state institutions have evolved from times of colonization and isolation to pass through centralization in the two-state era, reunification, decentralization and reorientation towards merit-based bureaucracy. Yemen’s reform experiences did not result in well-performing administrative and civil service institutions while they were often ad hoc, rushed or born out of political circumstances and donors’ pressure. Alongside the devastating civil war, contemporary state-building challenges result from legacies of reform failures. Current problems are exacerbated by long-standing policies of patronage, but they are also a reflection of the difficult reality of dominance of tribal elites, lack of capacities, interference from regional powers, and disagreements about the future shape of the Yemeni state.


PUBLIC INTEREST STATEMENT
Public administration in developing countries faces particular challenges. Often, administrative systems do not have a long history while the civil service sector is vulnerable to interference and political changes. To add to this, states such as Yemen might suffer from ill-defined issues such as institutional/political fragility and recurrent conflicts, which can hinder any capacity development. There is an important value in contextspecific measures to reform systems of public administration. In Yemen, the legacies of reforming these systems can reveal much about our ability to make improvements despite of a high degree uncertainty. Interplay of actors (including tribal and traditional forces), reform policies, politics and the overall socio-economic context can determine the faith of the public administration sector. Yemen's public administration witnessed different eras of post-colonial independence, separation (South and North), reunification, patronage, reforms and postrevolutionary conflicts. Lessons-learnt generate insights on administrative reforms in fragile states, state-building efforts and post-conflict reconstruction.
from legacies of reform failures. Current problems are exacerbated by long-standing policies of patronage, but they are also a reflection of the difficult reality of dominance of tribal elites, lack of capacities, interference from regional powers, and disagreements about the future shape of the Yemeni state.

Introduction
Public administration in the relatively young modern state of Yemen has been weak. Moreover, it has recently been facing additional, significant hurdles related to instability and (internal) conflicts. Even before the current-armed conflict, state fragility, ineffective and poorly performing institutions, and malfunctions such as corruption represented important challenges for public administration. As a result, civil service reforms as well as administrative reforms on the operational level have largely been intertwined with the performance of the underlying administrative systems and the outcomes of the overall political system. In fact, this is not surprising as the political-administrative context or the larger socio-political and economic context are often a determinant for the success of civil service reforms even in well developed countries. For example, Rouban (2015) points out a transformation of the political-administrative relations in European countries after the fiscal crisis of 2008 along the lines of a new subordination of civil service (to political and administrative elites) and a politicization of civil servants. In the same line, Peters and Pierre (2015) argue that civil service systems in developed countries have changed considerably due to post-bureaucratic concepts of public management such as those of New Public Administration and the concept of governance. The latter entails considerations of the civil service relationships to other actors such as the private sector, the public service recipients or the actors to whom the civil service system is accountable. Other contextual success factors for civil sector reforms include the strength of leadership, the (limited) role of donors and the scope of the reforms (Elayah, 2014;Polidano, 2001). A nuanced view of the relationship between civil service reforms and the overall administrative context is provided by Raadschelders and Bemelmans-Videc (2015) who analysed political and administrative reforms on the constitutional (e.g. change of regime and overall politicaladministrative system), collective (political system as well as the process and structure of administrative and bureaucratic systems) and the operational levels (political actors and civil servants). Accordingly, these three types of reforms are intertwined and can happen simultaneously, while system-wide reforms as well as reforms of politics and politicians are rare in Western (developed) countries. This paper links civil service change to state-building legacies in the modern history of Yemen. However, such reforms can be generally difficult to study in some contexts of developing countries. On the broader level, studies on reforms in public sector management in developing countries need to tackle some inherent problems such as measurement of results, contextual fits, uncertainty in reform process, role of donors and the political economy of targeted institutions and regimes (Brinkerhoff & Brinkerhoff, 2015). In fact, public administration reforms in some developing regions, e.g. African countries, have been radical and are often delayed until the occurrence of a fiscal crisis or a noticeable deterioration of public services (Batley & Larbi, 2004). Further, many bureaucracies in developing countries exhibit specific problems such as the spread of corruption which is closely related to the lack of meritocratic recruitment (Dahlström et al., 2012). As for the nature and behaviour of civil service institutions in developing countries, research reveals a wide variation. Civil servants' attitudes vary across and within institutions and countries depending on issues such as depolitisation of civil service, nepotism fight, motivation and performance incentives, while good and bad institutions can co-exist or be only gradually (not radically) different (Meyer-Sahling et al., 2018). Regional studies on administrative reforms prove the difficulty to capture these nuances as there are important country differences (Adamolekun & Olowu, 2015;Cheung, 2005). At the same time, commonalities are rarely spatial but rather related to political-economic, historic or social factors (e.g. economic development, legacies of independence or colonization, political stability or state fragility, leadership etc.).
State-building in Yemen is still underway, and public institutions have been through much turmoil and various transitions since independence. After the isolation and domination period under the British and Imamate rules, centralisation was predominant in the two-state era, and since the 1990s, decentralisation and liberalisation efforts followed. Federalism and regional institutions are currently under debate as to whether they are the right choice for the future of Yemen in the post-war period. This paper aims at examining administrative and civil service reform legacies in Yemen and linking these legacies to current challenges of state-building and post-conflict reconstruction. It uses descriptive-narrative analysis by highlighting and modelling the different phases of reforms in terms of motivation, actions, actors and outcomes. Such a case study analysis enriches the academic literature on public administration reforms in developing countries under unique settings such as political fragility, conflicts in state-building (and state reunification) processes, and the role of traditional actors.

Yemen: state fragility, non-state actors and conflicts
Beyond being a part of the developing countries group, some salient features of the case of administrative and civil service reforms in Yemen will be analysed in relation to the challenges of political instability and the particularity of Yemen's state-building legacies. These legacies includes the relative isolation of Northern regions prior to the establishment of the republic in the mid-20 th century, the reunification the state systems in the North and the South in early 1990s, the merge of ideologically two different bureaucracies, and the role of non-state actors (tribes). Yemen has long been a classic example of fragile states, e.g. (Carment et al., 2015;Hill, 2010), while additional contextual challenges related to state-society relations and state-building legacies make it a complex and interesting case. The Republic of Yemen was born in the 1990s after the reunification of the People's Democratic Republic of Yemen (PDRY) in the South and the Yemen Arab Republic (YAR) in the North. South Yemen gained independence in 1963 after more than a century of British colonial rule. The geographic area of the former South is larger in size, but accounts for only around one sixth of the population of today's Yemen. South Yemen became a socialist state with one-party rule and socially liberal institutions. In contrast, most of Yemen's population live in the fertile highlands of the former North Yemen, which gained independence in 1962 after centuries of isolation under the reactionary rule of a theocratic regime called an Imamate. Both countries effectively started to build their own institutions since the early 1970s after the British evacuated their posts in the South, and the civil war in the North ended. Until now, Yemen has been a largely rural society, with only 35% of the population living in urban centres (World Bank, 2019a). This is especially true for the tribal territories of the former North, while the majority of the population in the South live in urban coastal centres, as the rest of South is almost entirely desert.
In the 1970s, the two states of the YAR in the North and the PDRY in the South started to build their public institutions based on a centralised model of governance. However, political instabilities during the 1970s hindered critical developments, as this period was still overshadowed by the liberation struggles of the 1960s. In the 1960s, the PDRY was fighting for complete independence from British rule, while the YAR was involved in an armed civil war between supporters of the republic and loyalists to the old regime of the Imamate. The YAR succeeded in 1970 to win over royalists thanks to the help of the Egyptian military under the leftist regime of the former Egyptian President Gamal Abdel Nasser. It did, however, witness a period of power struggles and political assassinations, which ended in 1978 with the rise of the former president Ali Abdullah Saleh. The PDRY also witnessed major political rivalries within the governing leftist party of the National Liberation Front (NLF), which governed until 1978 under a twin-leadership system. The NFL adopted a Marxist-Leninist governance system with significant monetary support from the Soviet Union, which secured for itself military bases close to the Arabian sea, and hence the global trade routes. There were periods of armed struggle between the PDRY and the YAR in the late 1970s. During the 1980s in the PDRY, the NLF was transformed into the Yemeni Socialist Party (YSP) after the consolidation of power in the hands of a single leader, Ali Naser Mouhammad, between 1980 and 1986, and later, after an intra-party bloodbath, Ali Salem al-Baydh from 1986 until the reunification of the YAR and the PDRY in 1990.
While the PDRY adopted a communism-based governance system during the 1970s, and a more moderate system under the YSP during the 1980s, the YAR governance system was not ideologically oriented, and besides, the YAR experienced a larger degree of stability under the presidency of Ali Saleh. In 1982, Ali Saleh founded the General People's Congress (GPC) as an umbrella political organisation with Arab nationalist ideology and a focus on development. The GPC was effectively the only ruling party in the YAR until the reunification. It included politically influential tribesmen, and represented a gateway for the allocation of public jobs. In the pre-reunification phase, both Yemeni states were in a phase of institutionbuilding, and before the discovery of oil in the mid-1980s, were economically sustained by revenues from agriculture, fishery, remittances of expatriates, and international aid. Yemen was reunified in 1990, while a civil war over the reunification was fought between the South and the North in 1994. After the military triumph of the North in this war, Yemen has embarked on economic and civil service reforms with little success as to be explained later. Furthermore, the ramifications of 2011 revolution have uttered a period of political instability and ultimately a civil war since 2015.
Yemen has for a long time been one of the poorest countries in the Arab region. Being ranked 168th out of 188 countries in the 2016 Human Development Index, poverty eradication has been a prime goal of state policies since the country's independence. Poverty and basic needs represent now an even larger challenge due to the eruption of political conflicts in 2011. Since 2011, in the wave of the protests of the Arab Spring, the country entered into a transition period when the long-time president of Yemen, Ali Saleh, resigned in 2012, and his political party signed an agreement brokered by Saudi Arabia involving power-sharing with the opposition. The transition resulted in an UN-sponsored national dialogue, which led to a new constitution based on a federal governance system. However, the constitution draft was not adopted due to the overtaking of power by rebels (the Houthi movement) in late 2014, who originated from the upper North tribes, and had the support of troops loyal to the former President Ali Saleh. In 2015, Saudi Arabia, alongside other allying states, declared war on the Houthi movement. This movement has adopted slogans similar to those of Iranian-backed militias in Iraq or Lebanon (Winter, 2010). As a consequence of this war, a large portion of Yemen is now threatened by hunger and the spread of diseases such as cholera (Human Rights Watch, 2018). At the same time, the Saudi-led intervention has been sharply criticised for humanitarian violations (e.g. indiscriminate bombardment), while the Houthi rebels are also accused of war atrocities (e.g. indiscriminate shelling and child soldier recruitment) (Amnesty International, 2018). Now in 2020, the Houthis enjoy unchecked powers in many Northern governorates, particularly after killing their ally Ali Saleh and persecuting the members of his political party in late 2017.

Wicked problems or unspecific reforms? Study's method and theoretical framework
The civil service change and reforms in Yemen are contextualised in this paper by linking them to statebuilding legacies and salient features of the Yemen's history. The aim is to explain cumulative failures and problems of the civil service systems beyond the narrow focus on the performance of certain interventions and reform packages. As explained earlier, such a contextualisation is highly relevant for developing states in general and particularly for states suffering from a lack of capacities and political instability. In the literature on civil service and public administration reforms in developing countries, there are two seemingly competing approaches to explain contextual issues behind reform failures and adverse effects. First, some studies refer to macro-level contextual factors and highlight socio-political, economic and ideological features in the environment of the administrative and bureaucratic systems. They stress the unique and ill-defined nature of reform experiences. Some of these contextual factors include the role of economic and fiscal crises (Batley & Larbi, 2004;Rouban, 2015), the role of constitutional levels (Raadschelders & Bemelmans-Videc, 2015), or the social structure including the role of informal actors (Elayah et al., 2018). One of the most prominent macro-level contextual factors highlighted in administrative reforms in developing countries is state fragility. However, it remains difficult, if not impossible, to gauge causalities or predict the impacts of such factors on reforms due to the complex nature of the relationship between the two categories of variables. For example, the prolonged states of political instability in Yemen since the independence of its two parts (the North and the South) in the 1960s might allow for a characterization as a fragile state. At the same time, the list of syndromes for fragile states is extensive and includes a weak capacity of security provision, lack of rule of law or basic services, fractionalism, low economic performance, frequent political conflicts, and authoritarianism tendencies (Menkhaus, 2010b). However, the determinants of state fragility are broad categories, e.g. lack of meritocracy, volatility in the state revenues, small human resource base (Feeny et al., 2015). This makes the diagnosis of fragility elusive and ubiquitous, i.e. this diagnosis is applied to more than half of the world's countries in some listings (Menkhaus, 2010a). In fact, state fragility is widely accepted as a wicked problem, i.e. ill-defined with each case as essentially and uniquely complex. Therefore, such problems require a closer integration of analysis of state-building and political dynamics, while focusing on determining political capacity and encouraging leaders in such fragile states to address fragility (Menkhaus, 2010b). At the same time, iterative and good enough solutions are more suitable, with political will as a key requirement to build capacities of fragile states (Brinkerhoff, 2014). Other authors criticise the confusing, policy-oriented and state-centric nature of the concepts of state fragility and state failure, and they encourage the study of government institutions with a multidimensional and historically based, macro-level analysis of vulnerabilities in individual cases (Nay, 2013). In the same line, Brinkerhoff (2010) stresses the need for specificity in capacity building of failed or fragile states and demands indepth knowledge as well as the consideration of the specific country context. Second, meso-and micro-level contextual factors relate to factors in the immediate environment of the bureaucratic systems or the neglected dynamics of involved actors and issues in reforms. Here, the underlying suggestion is that the adopted reforms do not match the specific context of the country. This context can be identified and addressed. Such contextual factors can include considerations about the nature of the political or administrative systems as collective systems and the operation of the civil service in terms of involved actors (Raadschelders & Bemelmans-Videc, 2015). Besides, wide-spread corruption in administrative systems, conflicts among institutions or interferences from donors can hinder the development of civil service and administrative systems (Al-Saidi, 2018; Brinkerhoff & Brinkerhoff, 2015;Dahlström et al., 2012). In the case of Yemen, the role of some of these factors has been extensively highlighted by Elayah (2014;). Accordingly, past administrative reforms did not adequately reflect local knowledge and context-specificity, while international aid has been basically unspecific and ineffective. Remedies for this kind of contextual issues are largely directed towards improving the targeting and specificity of interventions. They are reflected in recommendations to strengthen domestic legitimacy (rather than promotion of democracy and project aid) (François & Sud, 2006), promote conflict mitigation (Al-Saidi, 2017a; Brinkerhoff, 2011), incorporate local participation and knowledge (Elayah, 2014;; support decentralised local authorities (Brinkerhoff & Johnson, 2009), and incrementally increase state involvement in regulating and guiding non-state service provision (Batley & Mcloughlin, 2010). This paper uses a historical narration of state-building legacies and civil service change to integrate the earlier-mentioned approaches in evaluating the role of contextual factors. By using a longer-term perspective, it shows the complementarity of these approaches in explaining cumulative developments of the civil service systems. Table 1 summarises the different state-building episodes and their implications on the development of civil and administrative systems in Yemen. The long-term implications represent cumulative effects that are carried out to the next state-building phase. Overall, the paper highlights the importance of examining and considering long-standing legacies that can determine the context of bureaucratic systems. It also advocates that public administration interventions should consider the overall political, historic and social contexts of a target case as well as the specific context of individual reform packages.

Centralisation of the two-state era
The first era of initial state-building in the North and South of Yemen has had important consequences with regard to the size, work ethics and politicisation of the civil service in modern-day Yemen.
Particularly, systematic changes in terms of nationalisations, unmeasured revenue spending, leadership changes and aid reliance have shaped the development path of the administrative systems. In the PDRY, according to Lackner (2017), thanks to the generous help of the Soviet Union (which was only reduced after Gorbachev's perestroika), the regime nationalised key sectors, leading to a flight of capitalists and a reliance on Soviets to build up limited industries. It also established national services, an education system with wide coverage, and a health system with the support of Cuba and China. At the same, there was an agricultural model of land collectivisation using small cooperatives, but also the unsuccessful experience of establishing state firms. The public sector and the military provided a large portion of employment opportunities (Lackner, 2017). At the same time, the progressive approach of the PDRY with regard to women's rights and gender issues was a famous one. The PDRY adopted one of the most liberal family codes in the Islamic world, and promoted women in pubic work to positions such as judges (Carapico, 2018;Molyneux, 1991). In contrast, the YAR had a rather conservative law system, which was later used as a basis for the new family law of the reunified Yemen (Dahlgren, 2013).
During the 1960s, the first efforts to build modern state institutions and administrative systems commenced. The YAR received some help from the Egyptian army to build some institutions to provide public services such as schools and hospitals. The entire administrative structure was modelled on the Egyptian model. It was only after the end of the civil war that a republican constitution was adopted in 1970, and during the 1970s, key institutions such as a central bank and specialised ministries were established. In general, according to Gable (1979), the performance of the administrative systems in the YAR suffered from constant interruption of leadership in the 1970s. At the same time, governmental revenues, which depended on taxation of agricultural products and trade, were thus limited and had to be supplemented by international aid  (Gable, 1979). In fact, the public sector constituted only a small part of the total economy in the early 1970s. Figure 1 provides some expenditure indicators in both the YAR and the PDRY.
During the 1970s and 1980s, remittances and migration to Gulf countries changed many dynamics, and have had ramifications for the administrative systems. Large migration of workers to Gulf countries experiencing a boom due to oil revenues led to an influx of money into the YAR. According to Fergany (1982), this resulted in high inflation, reducing the 27% nominal per capita GDP growth between 1970 and 1977 to only around 7% in real terms between 1970/71 and 1975/ 76. Furthermore, the remittances led to an increase in the YAR's dependence on aid, a severe shortage of skilled workers, an increase in imports and a trade deficit, a decline in the agricultural sector, and the neglect of local industrial development (Fergany, 1982). Although remittances started to decline by the end of the 1970s, they continued in the 1980s, and this led to further negative spillovers. According to Sultan (1993), remittances contributed to the spread of bureaucratic corruption in the 1980s as the enlarged, but poorly paid, public workforce resorted to corruption to increase its income. Here, with the government being more involved in development programmes using foreign exchange, and in regulating the growing business and urban sectors, both corruption and the public sector grew (Sultan, 1993). With increased revenues and an expanded taxation system in the 1980s, the government of the YAR was involved in developmental endeavours in order to provide public services. This is evident in a growing coverage with health services, infrastructural projects, and also an enlarged education system, see (El Mallakh, 1986;Peterson, 1984). The discovery of some limited reserves of oil in the YAR in 1984, and the start of its production in 1986, secured the country some additional income, which provided opportunities to increase the public sector further. This rise of mining activities is highlighted in Table 1 alongside other socio-economic comparison data.

Reunification and increasing dominance of Northern tribes
With the dissolution of the Soviet Union by the end of the 1980s, the now financially weakened PDRY entered into unification negotiations with the YAR. In 1990, a reunification into the Republic of Yemen (hereafter referred to as Yemen) was declared based on an agreement to share power and allow a multi-party political system. This significant change in the political systems, together with geopolitical events, greatly affected the performance and the structure of the administrative systems. The three political parties that emerged in the new democratic system were the YSP, the GPC, and a newly formed party, al-alIslah party, which represented a coalition of tribal and religious elements, many with ideological affiliation to the Muslim Brotherhood movement. After the reunification, the poorly performing civil service sector was supposed to be reformed. It was, however, left with little support in dealing with the burden of the sudden merger between institutions and public companies in the North and South. Moreover, the reform debates were interrupted by two events, the civil war of 1994 and the return of the many migrants ousted from Gulf countries during the 1990s.
The repatriation of most Yemeni migrants from the Gulf was due to the perceived pro-Saddam stance of the now-president of unified Yemen, Ali Saleh. This resulted in an escalation of prices and a heavy burden on state institutions. This mass repatriation (around 800.000 thousands including long-time residents) reduced YAR's foreign exchange receipts and exacerbated unemployment as well as the pressures on infrastructure and public services, although it did not constitute a disaster due to compensating effects such as the influx of investments by the returnees (van Hear, 1994). Moreover, the 3 years following the unification witnessed a growing frustration within the representatives of the former South with the political dominance of the North. The YSP obtained third place in the 1993 elections due to the higher population concentration in the North, and the issue of voting across tribal and regional affiliations (Detalle & Hiltermann, 1993). Southern representatives grew impatient due to perceived unjust treatment and interferences from Northern tribes, which were seen as more conservative and extraneous to the progressive socialist past of the South. Moreover, the discovery in the early 1990s of large fields of oil in territories of the former South, combined with an economic decline partly due the mass repatriation and a rise in corruption, ignited separatist aspirations (Hudson, 1995). As a result, upon a unilateral declaration by the leaders of the YSP of a renewed separation, a brief civil war ensued in the summer of 1994. President Ali Saleh was able to mobilise Northern tribes, together with religious forces, especially from al-Islah party, and won the war, effecting a complete dominance of the North. Ali Saleh enjoyed afterwards-until the Arab Spring of 2011-almost unchecked powers over state institutions and the military. At the same time, the effects of this forced unification were never solved, and are currently reflected in separatist movements in the southern regions. At the same time, the public sector emerged deeply weakened from the political turmoil of the 1990s, and in dire need for reforms.
Another ramification of this era is the influence increase of informal actors in the civil service systems, particularly the tribes. In fact, the alliance between the state and the tribes, particularly in the North, is not a phenomenon confined to the 1994 war. Tribal forces have been a dominant political force throughout the history of Yemen due to its social and climatic fabric. Yemen has a largely desert climate, with only around 2.5% of the country's total area considered as arable land, namely in highland areas and in valleys formed by small water streams, called wadis. In the mountainous areas, especially in the northern part, tribes have historically enjoyed a degree of autonomy. As a result, the government had limited influence outside urban centres and tribal mediation has been highly important for conflict resolution in Northern regions, see (Adra, 2011). At the same time, the development of tribes was different under the PDRY in the South, where the regime forcefully suppressed tribal identity in favour of socialist values. The PDRY sought to construct a society based on wide empowerment and social conduct, which is only partially based on (tribal) customs and traditions, see (Dahlgren, 2010). In contrast, the regime of Ali Saleh in the North used alliances with and favours to the tribes as a way of maintaining power. In fact, tribes became empowered during the 1980s by their participation in the new development efforts, and in remittance-based projects after the assassination of the popular President Ibrahim Al-Hamdi in 1977, who was perceived to have unfavourable opinions on the role of tribesmen in modern state institutions, see (Dresch, 1989). After the 1994 war, the influence of tribesmen grew even more as tribal repression subsided, and tribes became omnipresent for dispute resolution, creating social order and fulfilling public functions in the absence of the state (Maktary & Smith, 2017). Arguably, the influence of the tribes (e.g. the Houthi rebels) in the Yemeni conflicts of the early 21st century is very large, especially in areas north of and surrounding the capital city of Sana'a, see (Salmoni et al., 2010).
Before the current conflict, and under the Ali Saleh regime, Northern tribes from the Zaidi branch of Islam were overrepresented in the military and in public jobs of higher ranks. Around 30% of Yemenis are Zaidi Shi'a Muslims-a religious affiliation that resembles Sunnism rather than Shi'ism. However, this group is geographically confined to mountains around, and north of, Sana'a. Besides, many of these tribes have strong personal ties with, and receive funds from, the Saudi regime. In fact, the outside influence on Yemeni tribes is a major driver of their power, and a hindering factor in the development of modern state institutions, see (Peterson, 1984). In the civil service, the influence of informal, tribal actors has been studied by Elayah et al. (2018). Tribal and family affiliations have been the most important ties for the civil servants, while many civil servants have occupied offices using both primordial (family/kinship, tribal, regional and religious) and competing ties (e.g. through political parties or professional organizations). Besides, tribal actors were appointed to top and middle management levels of the civil service system in order to guarantee support of the ruling regime, and they, in turn, appointed lower-level personnel loyal to them (Elayah et al., 2018).
At the same time, the allocation of public funds and the distribution of public services under the Saleh regime were not always predominantly in favour of Zaidi tribes, to which Ali Saleh belongs. Former Southern governorates, as well as those in the middle highlands, where an affluent and less-tribal middle class is living, received a significant share of the new infrastructure and services. Still, politics was determined based on tribal alliances, while the military exhibited a clear familytribal structure, see (Fattah, 2010). As an example in the political sphere, the head of al-Islah party, the influential tribal chief Abdullah Bin Hussein Al-Ahmar, was the president of the parliament and a political ally of Ali Saleh until his death in 2007. However, al-Islah party was not the major force in the parliament, which was overwhelmingly controlled by the Saleh's GPC since the elections in 1997. In fact, it was officially considered to be the largest opposition party to Saleh, and besides, the GPC party evolved to include groups from the South and other regions. These groups were not connected by a clear political ideology, but instead, the GPC became more dominated by simple regionalism, tribalism and corruption (Al-Yemeni, 2003). In fact, tribes gained authority since the civil war, with the budget of the Agency of Tribal Affairs (a part of Internal Affairs Ministry) grew significantly (tripled between 2001 and 2012), while tribal patronage was an important element in the distribution of public jobs with many (one third) of civil service jobs as "ghost jobs", i.e. existing only in the payroll (World Bank, 2015).

Civil service, liberalisation and decentralisation reforms
After the civil war, it became increasingly evident to the Yemeni government and international donors that the public sector required restructuring and reform. Due to the decrease in workers' remittances, the rapid growth of population and the increase of inflation, as well as foreign debt, the government entered in 1995 into an agreement with the International Monetary Fund (IMF) to implement a macroeconomic adjustment and structural reform. This reform included the reduction of indirect subsidies, the liberalisation of the many interest rates, and the adoption of floating exchange rates for the currency. The economic impact of these reforms was generally positive, with an increase in GDP growth in the late 1990s and early 2000s. In fact, there have been fiscal improvements in terms of consolidating the budget and reducing the total public spending, which accounted for around 30% of GDP between 1996 and 2002, of which around 40% was for wages (Chemingui, 2007). Nonetheless, due to a higher public engagement in poverty eradication, public spending in social sectors increased from 11% of GDP from 1991-95 to 17% of GDP from 1996-2001 (Chemingui, 2007). After that, the total public spending continued to grow to around 40% of GDP in 2006, with subsidies, especially for fuel, accounting for, depending on international fuel prices, up to 10% of GDP during the 2000s (Al-Batuly et al., 2011). In fact, the new revenues from discoveries of moderate reserves of oil and gas during the 1990s helped the government to expand social welfare and sustain growth despite the increase in population size. Before the eruption of civil war in 2015, oil and gas earnings accounted for around 25% of GDP, and more than 60% of government revenues (International Monetary Fund, 2016). Still, the reserves were too limited to generate a strong and sustained development for the large Yemeni population, let alone prosperity at the level of the neighbouring Gulf countries. Figure 2 shows some key governance indicators for the period after the civil war of 1994, indicating a deterioration after this war, with a noticeable improvement during the late 90s and early 2000s, followed by a strong decline, particularly of political stability.
Alongside fiscal reforms, the government has embarked on various other reforms to improve the performance of public administration. In 1998, it launched a civil service modernisation programme with the World Bank in order to boost performance in the public services sector, improve budgeting and financial performance, curb corruption, and develop clear personnel management with a merit system. This programme was launched against the backdrop of civil service employment becoming a safety net for the unemployed or the poor while the selection was carried out on an ad hoc basis (World Bank, 2000). In 2000, the Ministry of Civil Service developed instructions on the recruitment procedures requiring announcements of all positions, and the establishment of a specified selection process. However, such new instructions did not remove barriers for merit-based recruitment as public service was not decoupled from political-administrative interferences (e.g. no entrance exams, appointments by decrees for senior management and by cabinet or minister for lower management positions and civil servants) (United Nations, 2004). According to the World Bank (2008a), there has been some initial success with regard to the provision of employment and payroll data, improving transparency on hiring, training programs, and providing computerised financial management systems. However, as can be seen in Table 2, the reforms had limited success in significantly downsizing the workforce, reforming the compensation system or reducing the budgetary burden. Since then, the World Bank has given some reasons for the modest success such the existence of low capacities and the overambitious agenda while it admitted that an incremental approach with realistic expectations might have been better (World Bank, 2008b). In fact, the civil service reforms in Yemen have been blamed for exacerbating poverty problems and acutely affecting educated females who depended on the civil service sector to ensure their job market participation (Colton, 2010). Still, the average size of government employment as a percentage of total working age population in Yemen is, with 22.5% as   Table 3. During the current war since 2015, the wage bill is consuming 2/3 of total expenditure, while revenues only cover 2/3 of this bill (World Bank, 2019b).
In addition, since 2005, Yemen implemented a set of reforms in order to improve good governance and fight corruption. For example, the judicial system was improved in order to attract investments and provide better judiciary services. In 2005, the cabinet approved a public financial management reform aimed at improving budgeting, procurement and financial accountability, and in 2007, the financial and accounting management systems were implemented in some ministries with the help of the World Bank. These reforms have led to tangible improvements in reporting and monitoring, as well as the increased digitalisation of processes. Importantly, between 2005 and 2010, Yemen adopted an action plan for improving good governance, which led to the establishment of an independent national anti-corruption commission, a procurement law, an anti-corruption law, a higher commission for procurement and tendering, and a higher authority for tender control. However, the results of these reforms were slow to materialise. In fact, during the 2000s and leading up to the Arab Spring of 2011, local corruption, patronage and resource capture remained evident, see (Alley, 2010). During the 2000s, an important reform process of decentralisation took place. In addition, a restructuring of the public productive sector was implemented with the aim of commercialisation and private sector participation. With regard to decentralisation, the 2000 Local Government Law initiated local administration reforms. The law, with its amendments up until 2008, gave the 21 governorates of Yemen greater administrative and financial autonomy by allowing direct election of a local council, which approves a governor and decides on social affairs, services, planning, development and finances. However, local councils could only collect taxes and fees, especially the Islamic tax called Zakat. The revenues from these sources were quite limited, and the issue of local governance remained controversial through political debates and elections until 2011, the reason being the perceived rise in corruption as a result of decentralisation.
Decentralisation was seen as a way to increase participation and democratisation in the wake of increasing frustration with the central government in some governorates in the former South and also in the North: e.g. in Sa'adah, where the Houthi rebels started several rebellions since the early 2000s. The decentralisation reforms were also implemented for some public utilities, and linked to local administration reforms. For example, in the urban water supply and sanitation sector, a comprehensive decentralisation reform was implemented through the establishment of local corporations for water and sanitation in many of the governorates. As a result, the former national water supplier lost its most important branches, and its future role became uncertain. Donors offered the independent public corporations significant support in order to adopt commercial practices. At the same time, the local governor became the head of the executive board of the utilities. In the end, decentralisation had mixed results. The performance of utilities did indeed increase, but so did interferences. In the Yemen case, water sector decentralisation based on administrative considerations ignored critical hydrological and financial considerations of the supply cost function of the utilities, see (Al-Saidi, 2015;Mewes, 2011). Finally, the participation of the private sector through public-private partnerships (PPPs) was promoted by donor organisations for the provision of public services and infrastructure. However, after initial studies and lengthy deliberations, PPP projects were developed in selected areas such as the water sector (Al-shareem et al., 2015;Sahooly, 2003), but this sector remained geared towards poverty alleviation (and less towards cost coverage or economic efficiency) (Al-Saidi, 2017b).

Federalism debate and the spillovers of the civil war
The donor-driven reforms in Yemen have had limited success to rein in spending, professionalise the administrative systems and kick off development. At the same time, donor reliance increased significantly. It amounted to 18 USD billion between 1995 and 2012 while foreign aid become instrumental in developmental policies (Elayah, 2016). Over the years and despite efforts to reform state institutions, the socio-economic challenges grew while political dissent in Yemen increased, see (Boucek & Ottaway, 2010). The population grew from around 15 million at the reunification year to reach almost 25 million in 2012. After the reforms, poverty was reduced slightly from 40% of the population below the national poverty line in 1998 to around 35% in 2006, but with no significant change in rural areas (Pournik & Abu-Ismail, 2011). In 2014, immediately before the recent civil war, the national poverty line stood at around 49% (World Bank, 2017). At the same time, this severely water-scarce country faced serious challenges concerning a rapid depletion of groundwater resources and increased local land and water conflicts, and this crisis is rapidly growing, see (Varisco, 2019). Further, since 2007, political dissent in the former South was transformed into an organised peaceful movement, the so-called Southern Movement (SM), which protested against unfavourable treatment of elites from the North. Above all, the SM demanded social benefits for members of the former southern military who were partially released from service after the 1994 Civil War. In the North, from 2004 until 2011, the Houthi rebel movement fought an insurgency in the northern governorate of Sa'adah. They cited mistreatment, lack of services, and negligence of the central government.
Under pressure from donors, the government continued to liberalise fuel prices and, between 2005 and 2010, implemented several rounds of price increases. After the overthrow of Ali Saleh in 2012, the transition government implemented further price increases on gasoline, diesel and kerosene. These decisions were politically unpopular, but seen as necessary by international donors citing high subsidies in the region, see (Verme, 2017). In 2014, the decision to remove diesel subsidies and increase prices by 60-90% affected many farmers, and was later used by the Houthi rebels as a justification for the overthrow of the government (Al-Weshali et al., 2015).
During the protests of 2011 and 2012, the Arab Spring brought about a period of relative instability, and a deterioration of the performance of public administration. The transition period of 2012-2014 did not improve this performance. According to the United Nations Development Programme (2015) primary services such as electricity, water and local government services deteriorated as a result of "environmental factors", related political instability and rule of law, but also overlaps between local and central government. The transition period aimed at drafting a new constitution through an UN-supported National Dialogue Conference (NDC). In early 2014, the NDC resulted in an agreement about transforming Yemen into a federal state of six regions (four in the North and two in the South). Each region comprised several states, which correspond to the governorates under the current administrative system. However, the NDC did not detail the division of powers and specific responsibilities at the federal, regional and state levels. Instead, a Constitution Drafting Committee was established with the task of outlining critical issues such as the division of revenues, as well as participation and decision-making rules, see (Al-Khali, 2014).
The NDC represented a milestone in the political reconciliation efforts in Yemen and was welcomed by the civil society (Elayah et al., 2020b). Although all parties signed the NDC agreement, both the Houthis and the GPC, still under the leadership of the then resigned president Ali Saleh, were increasingly questioning this deal. The two groups together represented the Zaidi political elite in Yemen. This elite might have feared the loss of its century-long political dominance as a result of the federal system. Furthermore, the Houthis complained that their envisioned region comprising mostly Ziadi tribes would not have access to the sea.
The overthrow of the state by the Houthis in late 2014, with the support of elements of Ali Saleh's old regime, began in Sana'a and spread quickly to all of Yemen, with the internationally recognised government fleeing to Saudi Arabia. As a result, Saudi Arabia led an international Arab coalition to drive out the Houthis. Between 2014 and now, 2020, the ensuing civil war between the Saudi coalition on the one hand, and the Houthis and the GPC on the other, led to a severe humanitarian situation. It also reshaped public services and state institutions, and led to a near collapse of the civil service systems. The embargo, which Saudi Arabia imposed on Houthi-held territories since the beginning of its intervention, led to a serious deterioration of public services and the threat of food insecurity for a large portion of Yemeni population, see (Stavrianakis, 2017). With Yemen becoming one of the major humanitarian crises in recent history (Human Rights Watch, 2018), there is a collapse of public institutions in many parts of the country, while selfhelp (e.g. use of solar panels, or rainwater harvesting techniques) has replaced some public services.
With regard to civil service and public administration institutions, some institutions are split (e.g. embassies under of the internationally recognised government), while local institutions are controlled by the respective local power. Other institutions are duplicated, e.g. the Central Bank in Sana'a was moved by the Hadi government to Aden, but still operates in the two cities. In fact, the challenges vary from one region to another. In northern parts of Yemen, the Houthis took over public institutions and assigned so-called "supervisors" affiliated to the militia, who oversee the work of public institutions. In reality, these supervisors intervene in decisionmaking on resource allocation and administrative issues, and are connected to the Houthi leadership (Hill, 2017, p. 266;Young, 2019). The GPC and Ali Saleh officially allied themselves with the Houthis after both had been targeted by the Saudi bombardment. They later formed a unified government with the Houthis. However, this government continued to be dominated by the "supervisors" as shadow administrators appointed by the Houthis. After some dissent between GPC and the Houthis, the joint government collapsed when the Houthis killed Ali Saleh in late 2017. As for the internationally recognised government based in Saudi Arabia, it continued to operate in liberalised areas, especially in the South. Some management-level workforce still operates from other countries such as Egypt under the direction of the exiled government. This government is responsible for paying the salaries of civil servants in the whole country through the Central Bank in Aden. However, most civil servants have received incomplete salaries while the Hadi governments is accusing the Houthis of appointing tens of thousands of new public workers (Browning, 2017). In the "liberated areas" (from the Houthis) in the South, government institutions have received support from Saudi Arabia and allied countries such as the United Arab Emirates (UAE). In fact, until now, elements of the Separation Movement, local militias, and UAE-or Saudi-backed forces are often exercising more control over public institutions than the weak exile government is.

Contemporary challenges, reconstruction and the way forward
The Yemeni case provides several contributions to the academic literature highlighting contextual factors of administrative and civil service reforms in developing countries. It reiterates the need for perspectives that incorporate cumulative developments and path dependence in administrative systems. At the same time, it shows that integrating the totality of macro-, meso-, and micro-level contextual factors requires longer-term and historically-grounded analyses. On the one hand, Yemen is exemplary for the syndromes mentioned in the literature of administrative reforms in developing and fragile nations, e.g. rushed and overambitious reforms once a country is on the brink of collapse (Adamolekun & Olowu, 2015;Batley & Larbi, 2004), donor-led policies (Elayah, 2014), lack of consideration of developing reform capacities (Brinkerhoff, 2010;Brinkerhoff & Brinkerhoff, 2015), or underestimation of the interlinks between higher levels of administration (e.g. constitutional, collective levels including the political system and the directions of the administrative as a whole) and lower ones (e.g. operative levels including the configuration of actors and civil servants) (Raadschelders & Bemelmans-Videc, 2015). On the other hand, the study of administrative and civil service reform legacies in Yemen throughout different phases highlights unique challenges or combinations of challenges facing future reforms. This section introduces some lessons-learnt about both technical administrative remedies and the overall special sociopolitical context of Yemen. It also provides some insights for the reconstruction phase.
First, the overall contexts of Yemen in terms of fragility, informal actors and political-economic changes makes ready-made solutions ineffective and perfect remedies illusive. Yemen's legacies illustrate a common theme in developing countries although the answer is not to seek even more ambitious and speedy reform packages. In fact, experiences from developing and fragile states emphasise the virtue of acting ahead of anticipated administrative crises, but also reiterate the need for tailored solutions and a realistic management of expectations. This means that real improvements arise from experimenting with deliberatively-decided and iterative reforms. Such reforms can target capacity development based on harmonization of the efforts by the state and donors using in-depth knowledge of issues and a careful selection of targets (Brinkerhoff, 2010). In Yemen, the specific socio-political and economic context is rendered by multiple factors highlighted in the reform legacies. These include the issues of recurrent territorial or resource-based conflicts, role of non-state or traditional actors, relative infancy of modern state organizations, ramifications of separation and colonization and the mounting environmental concerns (e.g. water scarcity). This special macro-level context requires acting ahead with multiple-tailored and experimental reforms at the same time. Often in the past, reforms did not anticipate the administrative problems that were accumulated over many decades. Instead, they constituted a (rather late) reaction that was soon outdated by changing realities, i.e. rising populations and dwindling fiscal or environmental (mainly freshwater) resources.
Second, there is a high need for context-specific administrative remedies that accommodate the immediate contexts of the civil service systems and its operational legacies. Some lessons apply for the Yemen case such as the need to prioritise good governance, depoliticise civil service, introduce merit-based civil systems, and exert a strong political will in implementing future reforms. The failures of past reforms in Yemen can be attributed to the miscarriage of these general principles. The use of civil service as a way to buy loyalties of political parties or tribal forces has been a common issue through the different historical periods in Yemen, and is even more pronounced in some periods such as during and after the civil war. It undermined professionalism of the civil service system which became intertwined with the seasonal political changes. In order to solve this, it is important to understand and contextualise historic derails and failures. Some studies have generated valuable knowledge on how to utter more context-specific reforms in Yemen. There are important (soft) factors such as a wide participation, capacity building, partnerships with key actors and higher-level civil servants, incorporation of local knowledge as well as the involvement of civil society (Elayah, 2017;Elayah et al., 2018;2020b).
Finally, better administrative reforms should be integrated within a wider reconstruction and reconciliation project. Above all, political reconciliation is highly needed to move Yemen beyond the current disintegration. The current war has led to a process of fractionalization that is counteractive to the notions of equity, inclusiveness and reconstruction of the post 2011 revolution era. At the same time, reconstruction needs to involve strong modes of national ownership and checks against foreign interference or politicization of reconstruction aid (Elayah et al., 2020;Hamid, 2018; The Project on Middle East Political Science (Ed.), 2018). This is in fact in line with the literature on public administration in post-conflict contexts which reiterate the need to reinstall legitimacy of state institutions (Brinkerhoff, 2007). In fact, beyond peace, reconciliation and legitimization, there are several feasible priorities requiring careful deliberation, e.g. fighting corruption, improving participation (e.g. of youth, women and civil societal actors), addressing labour and public sector reforms or finding long-term solutions to use inefficiencies of water and arable land (Brehony & Al-Sarhan, 2015;Elayah et al., 2020a). The national dialogue prior to the current conflict has sought to tackle these issues. However, this process did not move beyond general principles and powersharing rules (e.g. the federalism agreement) and was later interrupted by the infightings among political elites, particularly among those tribally structured Northern ones not satisfied with the outcomes of the NDC (Elayah et al., 2020a). In fact, integrating non-state, traditional actors in future reconciliation and reconstruction efforts is a genuine future challenge. On the one hand, it is advisable for the state to integrate knowledge and resources of these actors in future governance systems and the development of modern administrative institutions. On the other, state institutions, public administration, and civil service systems need to exhibit some protection and resistance powers (e.g. through political leadership and some level of institutionalization of good governance rules in higher administrative systems) to any instrumentalization towards personal or tribal gains.

Conclusions
Political, administrative and bureaucratic levels of government are inextricably linked in terms of institutional legacies and actors, and they form a unique and determinant context for public administration reforms in developing countries. Contextual analysis of administrative contexts in cases characterised by the complex problems of nascent state institutions, political fragility, nonstate actors in traditional societies, and recurrent conflicts is highly recommended before prescribing any bureaucratic or technical reform packages. In Yemen, public administration has been suffering from performance problems and an unstable political environment ever since the country's independence. The legacies of isolation or foreign dependence, the resource imbalances in the wider Gulf region, the changing political ideologies in the North and South, and the mismanagement of public funds are some of the factors behind this. The reunification of Yemen meant consolidating organisations with different problems and operating in different contexts. Furthermore, the role of tribesmen in state institutions, and the tensions between tribal allegiances and the modern state were not addressed in the past. There is a need for local and feasible solutions in order to achieve peaceful cooperation between tribal and public institutions. At the same time, the misuse of tribal identity for institutionalised political power was an important obstacle of the modernisation and development project in Yemen.
Public administration literature on developing countries reveals much on the importance of demonstrating political will, taming the role of donors, curbing corruption, promoting meritbased recruitment or managing expectations of stakeholders. Often, reforms come at the last minute, and then lay out an overambitious and a radical action package. Such insights are equally valid for the Yemeni case. The reforms initiated in the 1990s came after the realization of decades-long problems of low performance, corruption, lack of commercial thinking, missing capacities and unnecessary regulations. The reform package included fiscal reforms, good governance legislations, local governance, commercialisation, reduction in subsidies and an ambitious civil service modernization project. It had mixed results in terms of the overall contribution to growth and the modernisation of public administration. The reforms improved fiscal accounting and, initially, reduced poverty and deficit. The recruitment of civil service became more transparent and the size of workforce was reduced. Later, the reforms were offset by political campaign promises, increasing dominance of tribal actors, and the outbreak of political turmoil. At the same time, recurrent problems of Yemen in terms of lack of industrial development and revenues, inconsistent strategies, lack of capacities, and resource capture by elites meant that the reforms were not appreciable by the large and increasingly poor population, especially in rural areas.
In fragile and conflict-ridden states, the academic literature stresses the importance of incremental and iterative solutions that consider political dynamics, mitigate conflicts and address missing capacities. In Yemen, the political turmoil since 2011 is a direct reaction to the failures of the state to provide development opportunities and effective institutions of public administration. For some people, the comprehensive process of the national dialogue offered the right opening for reforming the Yemeni state on the constitutional level and the level of the collective, administrative system. In fact, many Yemen's unresolved problems such as the fate of regional separatist aspirations, involvement of youth and women, role of tribal elites or the need for civil and meritocratic bureaucracies were tackled during this dialogue. Federalism was formally agreed on. However, it remains a polemic issue today and a rallying point during the civil war for some Northern elites who fear the loss of their centuries-old dominance. The federalism agreement did not solve the underlying and difficult questions related to the participation, revenue distributions and power-sharing rules. These questions entails long-entrenched politics and complex legacies of centralised and patronage-based administration. They also reflect modern-day challenges of a growing and young population in a resource-scarce country with mounting environmental challenges (e.g. water scarcity). In the meantime, Yemen is on the brink of state collapse, and its public administration institutions are in poor shape. Alongside internal pressures, it has been negatively affected by rivalry between regional superpowers and, since 2015, the inconsiderate use of military force in a predominately traditional and poor society. The immediate end of the current war can save the rest of the state and its people. If not resolved soon, this war is expected to lead to further deterioration of the remaining elements of public services, and ultimately to an even more desperate humanitarian crisis. Finally, there is a (moral) responsibility on the part of the international community to continue to support Yemen. This support in rebuilding the economy and infrastructure, as well as in continuing the reform process, needs to be oriented towards the long term, conditioned on good governance, and based on lessons learnt from past experiences as well as the country's own reform legacies.