Challenges to Decentralization of Disaster Management in Turkey: The Role of Political-Administrative Context

ABSTRACT

Global disaster policy frameworks promote decentralization as a prerequisite of good disaster governance. Using 44 interviews, this study contributes to the literature that focuses on disasters and decentralization by investigating three systemic mechanisms that seemingly challenge decentralization of disaster management in Turkey: introduction of oversight systems, resource-allocation failure, and central–local collaboration. The results indicate that these mechanisms are enabled by a combination of political-administrative system characteristics and disaster-induced processes. To better understand how to secure the benefits of decentralization, we should engage with a wide range of disaster management actor perspectives, and integrate research on political-administrative systems and collaborative governance.

KEYWORDS:

• Decentralization
• decentralizing reforms
• disaster management
• political-administrative system
• collaboration
• Turkey

Decentralization has become the siren song in the fight against the consequences of natural disasters worldwide (Garschagen, 2015; Scott & Tarazona, 2011; United Nations Development Program [UNDP], 2015; United Nations International Strategy for Disaster Risk Reduction [UNISDR], 2005). Decentralization is believed to contribute to good disaster governance by increasing local capacity and by bringing in local perspectives and knowledge through local actor participation. Decentralization is further believed to advance disaster management activities as disasters and disaster risks manifest themselves locally. Similarly, the activities of local governments and nongovernmental actors are believed to facilitate context-specific risk management solutions that are custom-tailored to the specific needs, wants, and capabilities of local communities (Garschagen, 2015). Responding to disasters and reducing their risks thereby require local capacity, both within and outside local governments (UNDP, 2015). Decentralized systems also “prepare for and respond to disasters more effectively relative to more centralized systems” (Ainuddin, Aldrich, Routray, Ainuddin, & Achkazai, 2013, p. 51). Yet, managing disasters effectively also requires multilevel collaboration (Derthick, 2007; Waugh & Streib, 2006), but such collaboration does not follow automatically from decentralization reforms, despite their hypothesized effects of improving the inclusion of local perspectives and actor participation.

Decentralization is a key ingredient in mainstream disaster risk reduction paradigms. Both the Hyogo Framework for Action (UNISDR, 2005) and its successor, the Sendai Framework for Disaster Risk Reduction (United Nations [UN], 2015)¹ , advocate the empowering of local authorities. At the same time, the wider decentralization literature shows that decentralization efforts may have unforeseen effects,yet research detailing how recent decentralizing reforms play out in various disaster management systems lags behind policy circles’ advocacy. The wider decentralization literature has, for example, established that decentralization attempts may lead to recentralizing backlashes (Haase & Antoun, 2015; Jesse, Agrawal, & Larson, 2006). This article investigates such a backlash within the Turkish disaster management system.

Global disaster policy frameworks also emphasize that national governments must take the lead in disaster policy to ensure that the issues are provided with sufficient funds and attention. This points to the need for collaboration between actors at different levels in the disaster management system. Hutchcroft (2001) asserts that decentralization reforms must be carefully attuned to “the preexisting character of central-local ties” to stand a chance of being properly implemented (p. 46). Furthermore, investigating states’ central–local (vertical) relations lets us understand the consequences of decentralization processes (Hutchcroft, 2001). Owing to the fact that disasters often overwhelm local authorities and call for central-level assistance, these vertical relationships are also critical for disaster management in and of themselves. To handle hazards and disasters more effectively, there is a “need for greater understanding of the linkages between national and local governance systems” (Miller & Douglass, 2015, p. 2). Studying decentralization processes is one way to fathom these linkages that can be both formal/de jure, concerning for example the political-administrative systems’ setup or resource allocations, and informal. The informal links may concern the quality of relationships or the extent of communication between system levels. Formal and informal linkages are also likely to affect each other. The setup of the political-administrative system for example gives rise to norms that in turn may affect the prerequisites for vertical collaboration. In this vein, Garschagen (2015) draws attention to the fact that the success of decentralization reforms largely depends on the overall political culture that manifests itself in the nature of state–society relations and decision-making structures.

Following the lessons from the unprecedented earthquakes in 1999, highly centralized Turkey initiated decentralizing reforms, partly spurred by the international aid organizations involved in the subsequent restructuring of the disaster management system. Centralized and reactive disaster management would now give way to local and proactive risk management. Yet, despite the ensuing de jure disaster management decentralization in 2004, 2005, and 2009, this article suggests that the de facto development of the disaster management system toward decentralization has halted and rather taken a centralizing turn in the wake of two earthquakes in 2011. That a centralized state recentralizes after decentralization attempts is neither surprising nor new (c.f. Derthick, 2007). Yet, by drawing upon 44 semi-structured interviews, this article aims to contribute to a deeper understanding of how this recentralization occurred. More specifically, this article investigates three essential mechanisms that presumably challenge decentralization processes: central government’s introduction of new oversight systems, central government’s failure to match local authorities’ increased disaster responsibilities with increased funds, and the extent of central–local collaboration. The article further discusses how these mechanisms are enabled by the Turkish political-administrative system. In addition, it is explored how disasters’ attributes may challenge decentralization processes within disaster management policy.

Decentralization in the literature

The decentralization concept

In general, decentralization involves “authority being spread out from a smaller to a larger number of actors” as well as from a central authority to a less central authority (Pollitt, 2005, p. 373). Decentralization can be administrative, political, and fiscal. Administrative decentralization (de-concentration) occurs when responsibility for certain issues is delegated to local branches. These local branches are appointed by the central government and are thereby upward accountable (Treisman, 2007). Political decentralization involves not only the delegation of power but also the devolution of power and exclusive authority to lower tiers of the government. These lower tiers are downward accountable to the electorate (Ribot, 2002; Treisman, 2007). Fiscal decentralization involves to what extent lower tiers of the government can “define their own tax bases, set their own tax rates, and determine their own public spending” (Treisman, 2007, p. 25).

Arguments in favor of and against decentralization

Reviewing the concept of decentralization, Pollitt (2005) lists a number of general arguments in favor of administrative and political decentralization. One such argument is that it can reduce information overload, which speeds up decision making and makes the public sector more efficient. Another argument is that decisions made closer to the citizens are more responsive to citizens’ specific needs. Similarly, administrative decentralization can reduce political tension since it reduces central governments’ political intervention in local matters. It can also promote innovation as new ideas do not have to travel all the way up through the hierarchy to get approved. By being able to see the results of one’s efforts and not “just being a cog” in a large centralized bureaucracy, administrative decentralization can further improve the motivation of public employees (Haase & Antoun, 2015; Pollitt, 2005). Supporters of political decentralization further propose that it brings political power closer to the citizens and that politicians become “less remote, more visible, and more accountable” (Pollitt, 2005, p. 381). Political decentralization is also said to increase participation, both during and between elections (Pollitt, 2005).

Decentralization reforms can also have negative effects. Again drawing on Pollitt (2005), centralization makes it easier to retain a critical mass of experts. Moreover, autonomous local authorities may be more likely to treat citizens unequally, which can result in increased ethnical or political tension. Decentralization also increases fragmentation, which makes it harder to attain efficient coordination of policies and programs. Decentralization further brings more complex accountability systems, which may create opportunities for blame shifting and/or avoidance. Research on natural resource management and public service delivery indicates that decentralization does not necessarily bring more effective service delivery (Bardhan, 2002; Hutchcroft, 2001; Jesse et al., 2006; Knill, 1999; Ribot, 2002; Rumbach, 2015). Decentralization can, for example, lead to increased corruption and decreased health and education services (Treisman, 2000). Ineffective and/or dysfunctional decentralization processes have been observed in a variety of political-administrative system contexts, like the Middle East (Haase & Antoun, 2015), South Asia and West Africa (Agrawal & Ribot, 1999), and Latin America (Jesse et al., 2006). In some of these “unsuccessful” cases, more time may have been needed to be able to reap the benefits of decentralization (Ribot, 2002), but most studies indicate that the right conditions for successful decentralization were nonexistent from the outset (Hutchcroft, 2001). The potential of decentralization reforms in general is hypothesized to depend on the institutional context (Bardhan, 2002; Hutchcroft, 2001; Knill, 1999; Pollitt, 2005) and the overall political culture and circumstances in which they are embedded (Garschagen, 2015; Haase & Antoun, 2015; Scott & Tarazona, 2011). Some critical aspects to ponder before attempting to decentralize are whether local jurisdictions have the needed capacity and financial resources and whether they are prepared to assume the responsibilities handed to them (Haase & Antoun, 2015; Jesse et al., 2006). Other relevant context features include whether citizens are fully informed and the strength of monitoring, accounting systems, and accountability mechanisms (Bardhan, 2002). Scott and Tarazona (2011, p. vi) for example claim that decentralization’s potential to improve disaster risk reduction is often not realized “due to low levels of citizen awareness and general barriers to participation” (at least in developing countries), which suggests some overlap between general context features and those that matter for disaster management.

If some of these conditions are absent when initiating decentralization reforms, problems may occur locally, which may cause a centralization “backlash” as central governments undertake countermeasures to address these problems (Haase & Antoun, 2015). Rumbach (2015) further notes that gaps between de jure and de facto decentralization may affect the feasibility of such reforms within the disaster management sector. These gaps often emerge when central governments launch “decentralization” initiatives to appease international donor agencies or domestic constituents (Agrawal & Ribot, 1999; Garschagen, 2015), while being reluctant to give up power (Jesse et al., 2006).

Many of the general arguments favoring decentralization are assumed to be valid also for disaster management. Decentralization is, for example, believed to enhance participation, capacity, communication, and coordination between sectors and levels of government. These features are also believed to have a bearing on vertical disaster management collaboration. Increasing local capacity is key as several localized services (including fire services, the police, critical infrastructure investment, and building code enforcement) are critical in reducing casualties (Toya & Skidmore, 2013). Decentralization may also increase local disaster governance capacity (Rumbach, 2015), facilitate preparedness activities, and increase public participation in disaster planning by incorporating local knowledge and increasing local control over resource spending (Escaleras & Register, 2012; Garschagen, 2015). National governments are, however, key in providing leadership and ensuring that disaster issues get sufficient financial resources and an increased political profile (Global Assessment Report on Disaster Risk Reduction [GAR], 2009), which calls for close central–local relations and vertical collaboration between actors involved in disaster management.

Within the disaster and emergency management literature, a rich body of research has developed, describing the role and function of various levels in disaster management systems, while also analyzing the implications of placing the main burden for managing crises and disasters centrally, regionally, locally, or a combination of the three, as often is the case (see for example Birkland & Waterman, 2008; Cigler, 2007; Col, 2007; Derthick, 2007; Farazmand, 2016; Lester & Krejci, 2007; Waugh, 1994, 2007). Seeing that intergovernmental coordination between various jurisdictions is pivotal to avoid fragmentation, which hampers disaster response, many studies have also discussed this aspect (c.f. Bae, Joo, & Won, 2015; Caruson & MacManus, 2006; Comfort, 2007; Kapucu, Arslan, & Collins, 2010; Kettl, 2003; Morris, Morris, & Jones, 2007). Efforts have also been made to shed light on how decentralized structures and systems impact certain aspects of disaster response, like information flow (Koehler, Kress, & Miller, 2014) and evacuation (Waugh & Ruback, 2014). Rather than focusing on a disaster management system in place, its degree of centralization or decentralization, and how it affects disaster response, this article is concerned with how structures and processes within the political-administrative system that challenge decentralization enabled a recentralization of disaster management following the 2011 earthquakes.

Finally, it should be mentioned that decentralizing policy has been extensively studied from the perspective of various community-based disaster programs, yet these initiatives are often developed or supported by nongovernmental organizations (NGOs) (c.f. Jones, Manyena, & Walsh, 2015), while this study mainly concerns de- and recentralization within the state.

Challenges to decentralization

Considering the role of political-administrative systems for decentralization in other policy areas, Pollitt and Summa (1997, p. 15) found that the characteristics of states’ political and administrative systems were what “most significantly influenced what was possible in terms of the scope, process, and speed of [new public management] reform”. Sozen and Shaw (2002, p. 475) studied the implementation of administrative reforms in Turkey and concluded that “the cultural factors embedded in the form of public administration” in a country must be taken into account when considering the likelihood of reform implementation (see also Pollitt & Bouckaert, 2011). Examples of cultural factors that challenge reform implementation in Turkey are longstanding values and attitudes within the administrative system that manifest themselves in centralized decision-making structures, seniority systems, and nepotism (Sozen & Shaw, 2002).

Jesse and colleagues (2006) compared the effectiveness² of six forestry sector decentralization reforms and found that the intent of the decentralization reforms was rarely implemented. “The political dynamics related to policy reforms play a crucial debilitating role in the divergence between the rhetorical claims for decentralization and the institutional changes that actually take place” (Jesse et al., 2006, p. 1877). They also uncovered prevalent mechanisms that central governments use to limit decentralization reforms and the power that local actors thereby gain. Two of these mechanisms will be investigated in this study: introduction of new oversight systems and failure to match local authorities’ increased responsibility with increased resources. As existing decentralization research underlines the critical role that central–local relations play in decentralization (Hutchcroft, 2001), the extent of central–local collaboration and its potential impact on decentralization processes will also be explored.

Method and data collection

It is empirically valuable to relate the study of decentralization reforms to recent disasters. Disasters give rise to various processes that make the role of the political-administrative system visible. In addition, echoing White (2011, p. 7), “The time- and geography-bound context of disasters in countries may be an excellent ecosystem in which to study the relationship between central and subnational units”. Following disasters, the involved actors further tend to reflect upon the workings of the disaster management system, which in turn facilitates the investigation of actor perceptions of that system (Phillips, 2014).

Decentralization research has mainly focused on outcomes of reforms, like service delivery, and less on implications of the reforms with regard to improving governance. This is partly due to less-commonly available data on, for example, the quality of participation and accountability (Fauget, 2014). White (2011) similarly mentions that the quality of relationships between different tiers of government is “difficult to examine over time, particularly when observing state institutions from the ‘outside’” (p. 3). de Vries (2000, p. 148) further claims that an “empirical base”, partly constituted by the opinions of the actors who are in the midst of decentralization reforms, is missing in large parts of the decentralization literature. This study’s interview material provides “inside” perspectives and thus captures such less-commonly available data, like perceptions of local capacity, central–local relations, and participation. Actors’ inside perspectives and opinions rarely get told (especially local actor opinions) and are often drowned by official documentation (Birkmann et al., 2010; de Vries, 2000).

Interviews were conducted in Istanbul, Ankara, and Van during 2013 and in Istanbul, Ankara, Van, and Erciş during 2015. Istanbul hosts many of the NGOs working in Van and Erciş after the earthquakes and Ankara hosts ministries, expert institutions, and the Disaster and Emergency Presidency’s (AFAD) headquarters (HQ). Interviewees were selected based on their experience in and of the Turkish disaster management system and/or their position during the 2011 Van and/or Erciş earthquakes. Turkish administration officials “are expected to not criticize the policies of government in effect, in order to protect their future careers” (Çiner, 2014, p. 446). Hence, all interviewees were anonymized in order to enable unguarded reflection and to respect the explicit wishes of some interviewees who requested to remain anonymous. This environment also motivated the use of snowball sampling on site where trusted intermediaries put me in contact with interviewees. The first wave of respondents was contacted through four interlocutors (one relative of an informant who is a well-connected teacher in Van, and three university employees: one Swedish disaster scholar, one Turkish disaster scholar, and one Turkish public administration/disaster scholar). Subsequent recommendations of respondents were based on their position in the disaster management system or experience from the Van and/or Erciş earthquakes.

Forty-four semi-structured interviews (lasting between one and three hours) were conducted in English and Turkish using local interpreters. The interviews targeted a number of wider topics related to the implementation of the reform, the creation of AFAD, the importance of local knowledge, and response operations experiences. These topics spurred interviewees to reflect on issues related to decentralization, like vertical collaboration, participation, and capacity. Thirty-nine interviews were recorded and transcribed. Five interviews were not recorded but were promptly dictated onto the recorder and transcribed (see Table 1). Five key interviewees were interviewed twice as they had changed positions, which led to new and richer information.

Table 1. Breakdown of interviewees.

Actor type	Number of interviewees	Detailed information
National-level ministry official	2 (1 unrecorded)	Directly involved in the response
Central-level AFAD official^a	6 (1 unrecorded)	4 on site during earthquakes
Senior AFAD consultant	1	Many years’ experience advising and working with AFAD
State institution official	1	Disaster management dept.
Disaster management expert	3	2 former disaster management state officials
Van CCC representative (appointed AFAD and provincial ministry officials)	7	Directly involved in managing the Van earthquake (1 former)
Provincial ministry officials	2	Directly involved in managing the Van earthquake
Erciş CCC representatives (5 district ministry officials^b)	8	Directly involved in managing the Erciş earthquake
Municipal officials^c during the earthquakes	5	Directly involved in managing the Van or Erciş earthquake
Muhtars (Van and Erciş^b)	4 (1 unrecorded)	On site during Van and Erciş earthquakes
Turkish NGO representative	8	On site during Van and Erciş earthquakes
Semi-NGO representative	5 (1 unrecorded)	2 local and 2 central officials on site during the earthquakes and 1 Ankara-based central official
Turkey-based international NGO representative	2 (1 unrecorded)	Many years’ experience working with AFAD and Turkish disaster management actors

^a Focus group with three officials.

^b Focus group with district officials and Erciş neighborhood muhtars.

^c Focus group with two municipal officials.

The actors’ accounts were repeatedly read and the analyzed text segments were categorized into themes (recentralization, resources, capacity, local knowledge, central–local relations, etc.). Transcripts were contrasted and compared with each other as well as with primary sources (i.e. AFAD strategy documents, plans, reports documenting Turkey’s implementation of the Hyogo Framework for Action), secondary sources (i.e. news articles, academic articles, UN reports), and field notes. It should however be noted that in Turkey, it is generally difficult to acquire documentation from public sources (Çaha, 2009; Kapucu & Palabıyık, 2008; Yalçındağ, 1997) as public officials and employees often see themselves as owners of the state (Çaha, 2009) and “almost every kind of information is treated as confidential” (Tosun & Timothy, 2001, p. 352). Subsequently, public documentation covering the 2011 Van and Erciş earthquakes was scarce and not readily available.

Case study context

Countries that are frequently hit by natural disasters often have political-administrative and disaster management systems characterized by centralism and hierarchy that does not mirror the international disaster risk reduction discourse of decentralization and collaboration. Centralism is also a key defining characteristic of the Turkish administrative and disaster management system (Ganapati, 2008; Kapucu, 2012). The central government consists of the central administration in Ankara and the provincial administrations, which are further divided into districts. These entities are governed by provincial and district governors, valis and kaymakams, which make and implement decisions on behalf of the central government (Ganapati, 2008). The provinces and districts are administratively decentralized (deconcentrated), whereas the local governments are politically decentralized and elected by popular vote.

The public sector in Turkey is generally ineffective, and the provincial and municipal levels need increased authority and capacities (Freedom House, 2008; Gül & Kiriş, 2015; Kapucu & Palabıyık, 2008). The deficits are most severe in southeastern Turkey, where local administration participation is also low. Such participation is irregular across Turkey and often hinges on the “locality and the personalities of the individuals with authority” (Freedom House, 2008, p. 19). A planned comprehensive decentralization of the public administration system was initiated at the turn of the millennium but was vetoed by President Sezer on constitutional grounds in 2004 (Freedom House, 2008; Göymen, 2006). Perceived to threaten the country’s unity, the decentralization debate is controversial and politically sensitive in Turkey (Göymen, 2006; Gül & Kiriş, 2015). Such fears are particularly present in Turkey’s southeast, where Van and Erciş are located, due to the longstanding conflict between the Turkish government and the Kurdistan Workers’ Party (PKK).

Following the devastating earthquakes in 1999, three out of the four types of local authorities existing at the time of the Van and Erciş earthquakes consequently saw an increase in their disaster responsibilities (special provincial administrations,³ municipalities, and metropolitan municipalities). The municipalities’ and the special provincial administrations’ disaster responsibilities were increased through changes in acts 5393/2005 and 5302/2005, respectively. These acts contain identical emergency planning tasks and hold the administrations responsible for providing locally adapted disaster and emergency plans that aim to reduce losses in fires, industrial accidents, earthquakes, and other natural disasters (Balamir, 2013; Keleş, 2013). Moreover, the metropolitan municipalities act (5216/2004) increased the power, duties, and responsibilities of metropolitan municipalities (Gül & Kiriş, 2015, p. 48) and holds them responsible for preparing metropolitan-level plans and other measures related to natural disasters, as well as for vacating and demolishing risky buildings (Balamir, 2013). According to a government report (Ministry of Public Works and Settlements [MoPWS], 2009, p. 7), the increase in local authorities’ tasks and responsibilities meant that “most of the [disaster] mitigation, preparedness, planning, and recovery works have been transferred to them”.

The disaster management system reforms following the 1999 earthquakes culminated in 2009 with the creation of the Disaster and Emergency Management Presidency (AFAD) under the Prime Ministry. Three agencies with disaster responsibilities were combined and AFAD became the sole disaster authority (AFAD, 2016). In addition to the HQ in Ankara, AFAD also has provincial and regional offices. At the time of the Van and Erciş earthquakes in 2011, AFAD’s provincial offices were attached to the provincial governor. It has long been stated that the Turkish disaster management system needs stronger local levels and improved actor collaboration (Ganapati, 2008; Kapucu, 2012; Unlu, Kapucu, & Sahin, 2010), and the reforms aimed to strengthen collaboration and make local authorities more powerful (AFAD, 2012). Decentralization was hence emphasized by changes made both in 2004–2005 and in 2009.

During disasters, extraordinary entitlements are given to valis and kaymakams who head the Crisis Coordination Centres (CCC). Deputies to the valis and kaymakams, together with provincial ministries, security/law enforcement services, and NGOs are represented in the CCC. Municipalities do not assume any particular responsibilities during disasters (but come under the authority of the governors) but are responsible for certain relevant services such as police, fire, and infrastructure (Kapucu & Palabıyık, 2008). Turkey’s recent focus on risk management also puts more emphasis on preparedness activities, which to a large extent fall on the municipalities.

Analyzing the implementation process of decentralization reforms

The tripartite analysis revolves around the mechanisms through which Turkey’s central government may limit decentralization. The first two parts are structured around the two previously identified mechanisms: introduction of new systems of oversight and failure to match local authorities’ increased responsibility with increased funds. The third part explores the extent of central–local collaboration in disaster-related issues and how it may serve to limit decentralization. The analysis will also consider how these mechanisms are enabled by the Turkish political-administrative system. In the Turkish system, there are actors located at the local level who belong to the central government and thus are “central”. In the analysis, “local actors” hence include both administratively and politically decentralized actors, i.e. representatives from the provincial and district governorates, from AFAD provincial offices, and from municipalities. The provincial and district actors are administratively decentralized and thereby upward accountable to those who appoint them. The municipal actors are politically decentralized and thereby downward accountable to their constituency.

Introduction of new systems of oversight

The sections below discuss AFAD’s location, mandate, and change in appointment and budget procedures that contributed to recentralizing the disaster management system. Interviewees conveyed a near-consensus regarding the centralization of disaster management following the 2011 Van and Erciş earthquakes. “It is more centralized now, AFAD has everything – local authorities [have] nothing” (interview 36). A district-level interviewee said that

The experiences from Van and Erciş indicated that the efforts at the local level are not enough, so it doesn’t seem to work at the local level actually. In my opinion, if you take a global look at what AFAD is doing, they are actually centralizing their approach now. (interview 34)

A change made in 2014 tied AFAD’s provincial offices closer to AFAD HQ, making them less dependent on the provincial governors (act 6525; AFAD, 2015). AFAD’s provincial managers are still responsible to the provincial governors, but the appointment procedure has changed; AFAD’s president and the deputy prime ministers, and no longer the provincial governors, have the upper hand in assigning AFAD’s provincial managers. One provincial interviewee aired the suspicion that this change was made to allow AFAD’s president to appoint individuals close to him to important positions (interview 35). Favoritism is widespread in the Turkish political-administrative system (Çaha, 2009; Kalayçıoğlu, 2001).

The budget procedure also changed; now the central AFAD’s budget covers the provincial AFAD offices and their activities (act 6525). When the provincial AFAD offices were covered by the governors’ budgets, AFAD HQ found it difficult to implement projects that they had initiated (interview 11, 19, 39). “They [provincial AFAD offices] have to make their plan together with the vali [the provincial governor], so if they are not in agreement, they cannot do anything. The vali is very strong in Turkey” (interview 11, district interviewee, interviewed before the change. See also Balamir, 2013). AFAD HQ officials shared that the new structure also entails that AFAD’s provincial managers have “the power of AFAD behind them” (interview 43) and that they do not have to rely on the provincial governors, which they claim reduces the risk of economic or political conflict between them and AFAD HQ (interview 16, 19, 43). A semi-NGO representative said that “AFAD faced many problems with the governors. AFAD writes the theory [plans] but the governors touch the ground. They know what works and what doesn’t, and they reject Ankara’s theory” (interview 42). AFAD HQ interviewees claimed that the centralizing change will increase capacity at the provincial level as the governors can no longer halt capacity development projects. They further stated that if provincial AFAD managers are underqualified to work with disaster issues “we can say that these people cannot do this job and we can assign experienced AFAD people to do that job” (interview 43).

AFAD’s creation in 2009 partly aimed to strengthen local authorities. Indeed, one aim of the disaster reforms was to achieve a more decentralized approach (AFAD, 2015). Yet, to avoid the inertia previously experienced, AFAD was located in the Prime Ministry, which was a move toward centralization. Previously, three different general directorates under different ministries dealt with disasters, but “this set-up did not fit the system” (interview 32). A former employee of the general directorate for disaster affairs said that being at the same hierarchical level made it “very difficult to give some, let’s say, suggestions or other commands to the other ministries. Because you know, they are ministries and you are a ministry” (interview 36). This interviewee further explained that one unit must be able to direct, push, and monitor other ministries’ activities, but in order to be able to do so, this unit must be located higher up in the hierarchy than the ones it should oversee (see also Japan International Cooperation Agency [JICA], 2004; Kapucu, 2012). An AFAD HQ official confirmed this by saying “because of our power, AFAD can trigger other ministries to do some duties” (interview 43).

Moreover, being located within the Prime Ministry would normally “only” grant AFAD with a regulation and coordination mandate. Implementing mandates resides with the general directorates within the ministries. However, AFAD HQ has gradually become involved in operations (interviews 23, 24, 36, and 39). In practice, it now has an implementation mandate. Some believe this arrangement is beneficial, while others see it is an interim solution necessitated by the lack of local-level capacity. While hoping that the solution is temporary, one international NGO representative maintained that the implementing mandate gives AFAD “good coordination power and all the support mechanisms it needs to infuse into different ministries” (interview 24).

Many interviewees from different system levels laud the changes made (interviews 4, 5, 6, 10, 11, 15, 16, 19, 29, 31, 36, and 37) saying that locating AFAD within the Prime Ministry has mitigated the persistent problem of system fragmentation. Others from various system levels are positive to certain functions being centralized, like the coordination of search and rescue and database maintenance (interviews 6, 13, 32, 34, and 35), but they are not in favor of a fully centralized disaster management system. A provincial official said, “Before, AFAD managed locally but they didn’t think this was working so now it has been centralized. Actually, it should have been the other way around” (interview 35). A district official held a similar view, saying that the local actors should be prioritized more “but I think they are kind of being ignored. I think it has something to do with Turkey’s own peculiar conditions” (interview 34).

A few central actors meant that local officials’ “openness to partisan pressures” (Keleş, 2013, p. 8; interviews 29 and 31) motivates having certain functions centralized, like damage assessment. With regard to maps showing the probable areas where households are affected by an earthquake (which lay the ground for state compensation), one AFAD HQ official explained that they can be manipulated to include certain areas that would not be included based on geological calculations: “If you have good political connections, these maps can be redrawn” (interviewee 19). Such accounts indicate the presence of favoritism.

Failure to match increased local responsibilities with increased funds

This section concerns the central government’s allocation of resources to local authorities’ increased disaster responsibilities. Local capacity issues and local actors’ hesitance toward having disaster authority delegated to them will also be discussed.

Turkey’s local levels lack disaster management capacities (Keleş, 2013; Unlu et al., 2010), which according to both central and local interviewees speaks in favor of centralization (interviewees 29, 31, 32, and 36). According to AFAD HQ and an international NGO representative, it also motivates AFAD’s operative approach and mandate (interviews 16, 19, and 24). Yet some central-level interviewees held that centralization causes and upholds the capacity deficit. “Everything should be decentralized, and knowledge should stay in the Van area and be institutionalized. […] When you say that all this information is centrally located [at AFAD], there is no demand for it locally” (interview 37). One former central official pointed out what others also referred to: that the longstanding tradition of having centralized structures in Turkey entails that the local actors are highly dependent on central actors and that they are unaccustomed to taking initiatives (interview 31). In the same vein: “There are provincial [local AFAD] offices but that doesn’t mean anything. There are just a few people sitting there waiting for orders from Ankara” (interview 37).

Provincial and district actors also cited capacity deficits as a reason for being unable to shoulder disaster responsibilities. They also felt unprepared and incapable of assuming the responsibility handed over to them (interview 10, 32). A district official said “What do I understand from crises and earthquakes, I am not a [disaster management] professional but they came and gave me the responsibility […] What can I do, this is not my area of expertise” (interview 32).

Interviewees across the board reasoned that the capacity of the local actors (the provincial governorates, the district governorates, AFAD’s provincial offices, and municipal levels) to deal with disasters is equally insufficient (interviews 1, 11, 12, 16, 19, 31, 33, 36, 43, and 44). The following quote from a former central official summarizes the opinions: “Disaster laws in Turkey give a lot of responsibility to governors and municipalities but they have no idea how to manage disasters” (interview 36). Interviewees mentioning the municipalities specifically pointed out that their increased responsibilities have not been matched by an increase in financial and human resources (interviewees 39 and 42; see also Kapucu, 2012). A semi-NGO representative said, “They try to decentralize maybe, but if you decentralize responsibilities, you must also give more financial resources” (interview 42). Yet in Turkey, fiscal decentralization generally remains very limited (EU, 2014). When combining this lack of ability and experience with the respect for rank in the system that makes it difficult to criticize high-status officials, the capacity issue becomes particularly palpable. This is captured in a quote from an NGO representative concerning provincial governors: “They learn disaster management during the disaster. And also, no one can challenge a governor, which means no one can tell him that he makes bad judgements” (interview 44).

The academic and practitioner disaster discourses alike emphasize the need for local ownership and empowerment. Yet, many local actors involved in managing the 2011 earthquakes rather expressed a wish to be relieved of their current disaster responsibilities. Provincial and district officials alike conveyed that they were themselves affected by the disaster and therefore partly unable to help others (interviews 8, 10, 13, 38, and 43). Many described their struggles of having a combined responsibility for taking care of their families and also for managing the earthquakes (interviews 5, 6, 8, 32, and 41). Suffering from bereavement, one provincial official said: “I waited for them [missing family members] by the debris for six days, and I also had to be at the crisis coordination centre. How can anyone be productive in this state of mind?” (Interview 6). Some district and provincial officials, however, claimed that they still managed to organize very quickly despite these difficulties (interviews 5, 13, and 26).

The local actors who wished to transfer responsibilities and authority for disaster management upward simultaneously underlined the importance of local knowledge for effective response and recovery operations (interviews 11, 17, 32, 33, and 38). The provincial governors who are responsible for managing disasters in Turkey are often new to the provinces where they work, and may therefore be unaccustomed to the local conditions (JICA, 2004). Interviewees did not suggest that central-level actors be in charge of managing disasters. The great value placed on locality-specific knowledge may explain this. Rather, interviewees preferred a regional solution, with well-prepared authorities from neighboring provinces that possess contextual knowledge. Provincial- and district-level interviewees further suggested that available local staff should work intermittently as consultants and guides to assist the neighboring authorities’ management during a crisis (interviews 10 and 38).

Extent of central-local collaboration

By their nature, disaster-related issues transcend single governmental jurisdictions, which calls for the integration of central and local actors. This section explores the extent of central–local collaboration in disaster-related issues.

Local disaster management capacity in Turkey suffers not only from inappropriate funds but also from weak relations between central and local actors. An AFAD (2015) report mentions that the central–local relations must be improved as they strengthen the capacity of the responsible local agencies. One AFAD HQ interviewee also testified to the lack of vertical collaboration (interview 19). Another AFAD HQ interviewee lamented the decrease in interaction with field actors since AFAD’s creation and further described how “local actors involved in managing the earthquakes in Van and Erciş offered to discuss their lessons learnt and experiences with AFAD but their offer was not accepted” (interview 39). This central–local gap in the disaster management system is also mentioned by the local actors, especially related to their experiences after the earthquakes. A provincial official said that “AFAD [HQ] never asked about our experiences, although we had many. They weren’t enthusiastic to work with us. Unfortunately, when this work [planning and lesson drawing] is done, experience is not a criterion that is valued, politics are” (interview 35). Here, “politics” referred to practices related to favoritism rather than merit-based ones. There were many similar accounts; this one is from the district level: “The first and only one who have asked us about anything is you” (interview 32). Key actors, both local and to a lesser extent central, strived to communicate their experiences to AFAD. One key district official who had met with responsible individuals at AFAD said:

But unfortunately there was no comprehensive effort to share this experience. As long as we do not have a request from AFAD, I personally do not think it is correct to intervene in the process and push to share my knowledge. (interview 34)

This statement also illustrates the respect for rank that constrains vertical communication, upholds gaps between levels, and complicates central-local collaboration. Some actors have become unmotivated to work with disaster issues because of this lack of interest from AFAD HQ, while others (e.g. provincial, district, and semi-NGO officials) instead have created horizontal informal forums where they exchange experiences and lessons with other actors (interviews 26, 32, 33, 34, 35, and 42).

Interviewees conveyed a picture of a disjointed system where the local actors found it hard to make their voices heard upward. A former central disaster management official said, “The disaster management system is very very low in, let’s say, community participation and there is also no participation of local authorities” (interview 36). This disjointed system manifests itself in the distance between local and central actors’ perceptions and interpretations of events. Some interviewees, both local and central, expressed criticism saying that local actors were excluded from the planning of the new residential areas for earthquake survivors (interviews 33 and 36). The quote below is from a provincial official.

All commands come from Ankara, and as they are managing the process, the people recommending where [the houses] were to be built […], unfortunately don’t know about our local conditions and did not analyse the cultural profile. They don’t know what kind of problems we have in different areas, or about the infrastructure problems […] Maybe they are experts, but they don’t know Van. (interview 33)

Including local actors in collaborative resettlement processes is crucial to avoid unnecessary suffering (Birkmann et al., 2010; Tercan, 2015). For earthquake reconstruction, legislative modifications in 2011 meant that the previous decentralization of housing and urban planning issues was abolished and that urban renewal and land-use planning were recentralized (Balamir, 2013; EU, 2012). An expert on Turkish disaster management said “TOKI [the Ministry of Housing] and the Ministry of Environment and Urbanization could overtake all functions and competences from the local authorities as far as planning, housing and urban transformation projects are concerned. They replaced the local authorities” (interviewee 29). Interviewees were also critical to that the apartment houses were not adapted to traditional needs and that they were remotely located, which limits access to both labor and social arenas (interviews 33 and 36; Tercan, 2015). Local interviewees thus saw the implementation of these projects as far from ideal. An AFAD HQ report, however, conveys that the reconstruction of Van is a “best practice” that currently serves as a model for the future (AFAD, 2015). The accounts from local and central actors regarding the earthquake response also differed and were said to be “worlds apart, with one version describing quick and adept disaster response and the other suggesting that Turkey has a long way to go in terms of disaster management” (Ayasun, 2011). Such divergence in perception and understanding further complicates collaboration.

The norm against upward criticism (related to the respect for rank) widens the central–local gap and decreases the potential for vertical collaboration. It also conceals local capacity development needs. A provincial official said: “We should explain that we do a successful job. We can’t say we had any problems even if we did. If we do, it will cause us problems. We try to make ourselves perfect” (interview 26). Another provincial actor spoke of a high-level meeting held in Van during the earthquake response where he was asked if they experienced any problems. Upon answering yes, they retorted: “How dare you raise your voice, they are deputy prime ministers!” (Interview 33). Former AFAD consultants further described that AFAD HQ have stopped hosting independent advisory meetings. Some said it was due to AFAD’s overcrowded agenda, while others said that AFAD was uninterested in hearing the criticisms put forward (interviews 29, 36, and 37).⁴ Accounts from AFAD HQ support the latter. “Everybody closes their mouths. When our head of department or deputy director or somebody else says we were successful, nobody can say ‘No, we were not’ […] We do not have a chance to criticize the system freely” (interview 39).

Discussion

This study made a contribution to the international disaster practitioner and policy community by shedding light on how decentralizing disaster management reforms played out in practice in Turkey and by uncovering that disasters may amplify the centralizing logic already present in some political-administrative systems. Relating the study of disaster management decentralization and implementation to an actual disaster further made it possible to unveil how disaster-induced response and recovery processes can challenge decentralization processes. Disasters claim more resources than what is locally available, which necessitates central resources and involvement. Such involvement was not unwelcome by the local actors during the Van and Erciş earthquakes; rather the opposite. Yet, the gaps between central and local actors in the political-administrative system limited their exchange and collaboration. The interpretations of the unfolding of disaster-related activities, the shortcomings, and the proposed solutions to address them were thereby dominated by central actors and void of critical or local actors’ perspectives. The centralized tradition in Turkey coupled with the key role that central actors like AFAD HQ play in disaster management legitimizes their dominance (especially with their implementation mandate). Challenging perspectives or standpoints that diverge from central actors’ view are rarely told or heard, partly due to the respect for rank within the system that upholds norms against upward criticism. The discrepancy regarding the reconstruction process is an illustrative example: AFAD labeled it “a model for the future”, while many of the local actors, who had no chance to voice their opinion, were critical toward the process and its outcome. Central actors’ domination of the post-disaster narratives both widen the central–local gap and reify and regenerate the central perspectives, which challenges decentralization.

The disjoint character of the political-administrative system, the identified respect for rank, and the norms against criticism and local participation also present challenges for central–local collaboration. This, in turn, makes reaping the proposed benefits of decentralization, like increased participation, difficult. Rather than encouraging central–local collaboration and fostering a common understanding, this setting risks feeding vicious circles of distrust that maintain central–local gaps, which in turn challenges decentralization. It should be noted that the 2011 earthquakes took place in a historically conflict-laden area, which may have put additional strain on central–local relations. Following the attempted coup d’état in July 2016 and ensuing purges, polarization in Turkey increased further. Although it is too early to tell how the system managing disasters has been affected by the post-coup developments, a few challenges to disaster management decentralization and collaboration can be observed. A large number of elected mayors and municipal representatives were removed from their offices after charges of various anti-state activities and replaced by state administrators (Tattersall, 2016). Such measures thwart decentralization and the devolution of power by making local government officials upward—rather than downward—accountable to the electorate. It also challenges collaboration as the legitimacy of the appointed officials can be questioned locally. Removing locally elected officials from their offices also hinder collaboration locally, as the appointed state administrators seldom come from the local area to which they are appointed, which entails that they often lack ties to the community. This state of affairs affects disaster management, and collaboration related to it negatively as state-appointed officials lack access to local social networks, resources, and knowledge—factors that have proven to be key for successful disaster management.

The measures related to the purge have also affected many sectors in society that bear on disaster management. The hasty reforms in the wake of the attempted coup, for example, left the security sector at a critical juncture (Gürcan & Gisclon, 2017). As the disaster management sector and the security sector overlap, it is not far-fetched to assume that the disaster management system has been affected too. Other societal sectors that feed into the disaster management system, like the media, the local authorities, and the civil service, have also been affected by the post-coup measures (EU, 2016). The civil society, which plays a large supporting role in disaster management, not least locally, has also been reduced. By the end of 2016, close to 1500 NGOs had been banned after being accused of anti-state activities (Çetingüleç, 2016). These developments create polarization and distrust, which endangers collaboration. A final observation is that the purges across the Turkish society will result in a loss of competence as central and local actors with valuable accumulated experience from managing disasters risk being removed from their positions. Building expertise takes time, especially for local actors who more rarely deals with disasters than central actors, and losing it naturally influences the disaster management system negatively not only in terms of decentralization.

Conclusion

Despite de jure decentralization of disaster management responsibilities in 2004, 2005, and 2009, de facto decentralization of Turkey’s disaster management system has halted and rather taken a centralizing turn after two earthquakes in 2011. This article aimed to highlight how decentralization of disaster management was challenged and how recentralization came about. It did so by investigating three essential mechanisms that challenge decentralization processes: central government’s introduction of new oversight systems, central government’s failure to match local authorities’ increased disaster responsibilities with increased funds, and the extent of central–local collaboration. The article further discussed how these mechanisms were enabled by the Turkish political-administrative system. Finally, it was also explored how disasters’ attributes may challenge the decentralization process.

First, the central government’s introduction of new oversight systems and failure to allocate sufficient resources in the case studied underscore the validity of Jesse et al.’s (2006) mechanisms, also in disaster management. The creation of AFAD was part of a reform package that aimed to strengthen local authorities, but its location in the highest echelons of government and its current mandate also made it a move toward centralization.⁵ AFAD also recentralized by changing appointing and budget procedures. The central government also failed to match local authorities’ increased disaster responsibilities with sufficient resources.⁶ To succeed with decentralization, previous research has claimed that local actors must have sufficient preparation, capacity, and financial resources (Haase & Antoun, 2015). In this study, a few interviewees suggested, against conventional wisdom, that AFAD’s centralization of budget and appointment procedures may contribute to this by improving professionalization and securing finances for local disaster capacity development projects. Hence, while the recentralization surely challenges decentralized ideals in the short run, they may also simultaneously contribute to increased local capacity and participation in the longer run (Hermansson, 2017). The timing of decentralization reform thus has to be carefully considered.

Second, this article suggests that to understand the trajectory of decentralization reforms, the attributes of the particular issue area must be considered. In disaster policy, resources and capacity may not be enough to make unwilling local actors willing to assume disaster responsibilities, as they, after having suffered from disasters’ consequences, questioned how they would be able to undertake such responsibilities. Arguably, local actors doubting their ability to cope with delegated responsibilities due to being affected themselves rarely occurs within, let’s say, education policy. Central actors often use local actors’ insufficient technical capacities as an argument against decentralization (Jesse et al., 2006). Yet here, perhaps due to disasters’ unique attributes, this argument was also put forward by the local actors themselves.^{7 ,8} The local actors who questioned their new-won responsibilities did not uniformly wish for a centralized system; many rather promoted a regionally based flexible solution.⁹ Yet, given the state’s overwhelming role in disaster management, central, rather than local, actors’ interpretations of disaster activities, the shortcomings, and the proposed solutions to address them dominate discussions.

Third, by exploring the extent of central–local collaboration and how it affects the potential for decentralization, this study contributed to the disaster management literature by showing that prerequisites for disaster management collaboration and decentralization of disaster management are coupled. While often lacking technical disaster capacities, many local actors (especially municipalities) have much locality-specific knowledge to offer (c.f. Miller & Douglass, 2015). In Turkey, this knowledge largely remains outside of the formal disaster management circuit, partly due to the disjointed political-administrative system that hinders central–local communication and collaboration.¹⁰ Problematic for collaboration are also the norms in place that generate a great respect for rank, which aggravates upward criticism and complicates local participation. This shows how cultural factors and the characteristics of the political-administrative system influenced the trajectory of the decentralization reforms and the recentralization process (c.f. Sozen & Shaw, 2002).

In sum, this article contributed to expanding the knowledge foundation that current disaster policy and advocacy is based on. This article also showed that this foundation as well as the knowledge about how disaster management decentralization (and recentralization) plays out in practice in various political-administrative systems, need to be further enhanced. Continued fieldwork-based case studies that include a wide range of actor experiences and the characteristics of the political-administrative systems surrounding them may be particularly useful in this pursuit. Such studies may also suggest alternative strategies for improving disaster governance. This article indicated that the potential of decentralizing disaster management systems is coupled with the prerequisites for central–local disaster management collaboration. Future research into disaster management decentralization would hence benefit from engaging with the collaborative governance literature to consider how shared understanding, trust, and communication between actors at different levels may be nurtured. Achieving a fertile ground for central–local collaboration would not only advance disaster management, but would in all probability also help pave the way for reaping the benefits of potential future decentralization of disaster management.

Additional information

Funding

This research was carried out within the Center for Natural Disaster Science (CNDS) research school, supported by the Swedish Government strategic research funding to Karlstad University, Swedish Defence University, and Uppsala University.

Notes

¹ The frameworks outline the necessary actions from a multitude of actors to reduce disaster-related loss.

² Effectiveness entails to what extent decentralization reforms are actually implemented, not whether decentralization reforms lead to more effective outcomes.

³ Special provincial administrations “complement[ed] the municipalities’ mandates by providing certain services at the local level” but were removed after regulations put in effect in 2012 (The tenth development plan, 2014, p. 131) (Freedom House, 2008, p. 19).

⁴ Turkey and AFAD, however, asked the European Commission to peer review its Disaster Risk Management system in 2015. The full report is available at: http://www.falck.nl/nl/peerreviews/publications/Documents/ Peer%20Review%20Report%20Turkey%20EN.pdf.

⁵ See Bae et al. (2015) for a similar development in South Korea.

⁶ See Garschagen (2015) for a similar development after disaster management decentralization in Vietnam.

⁷ Not all actors representing the same administrative level shared opinions; all central actors did not prefer centralization and not all local actors were against it.

⁸ Previous studies of disaster governance convey a similar picture (c.f. Rumbach, 2015; Waugh, 2007).

⁹ Similar to the Emergency Management Assistance Compact in the U.S. See also Aoki’s, 2015, description of how a similar system emerged in Japan following the multiple disasters that struck the country in 2011.

¹⁰ See Rumbach (2015) for an account of how insufficient local resources and capacity hindered potential collaborative and decentralized disaster management in India (see also Bae et al., 2015).