Disasters naturally require decentralised decision-making and intensive human interactions (Kapucu & Van Wart 2008; Kirschenbaum 2004; Mileti 1999). Managing disasters involves dynamic processes that are ideal yet demanding. Thus, collaboration among organisations and government agencies is essential for the development of an effective strategy and better performance during disasters.

During disasters, government institutions are expected to have (Kapucu & Van Wart 2008):

Aside from the fact that collaboration often occurs among proximal and like agencies (Simo & Bies 2007), collaborative disaster management faces various challenges which oftentimes lead to the failure of the response operations. Poor communication, inadequate planning, misguided and poorly executed leadership, and insufficient coordination with various stakeholders lead to collaborative failures (Kettl & Walters 2005; Menzel et al. 2006; Wise 2006).

The Philippines’ geographical location makes it susceptible to the most disastrous cyclones in the region and has experienced its own share of collaborative failures. In response to the devastating social disruptions caused by typhoons, institutional mechanisms, such as the Republic Act 10121 (2009) or the Philippine law on disaster management, were implemented in 2010. The law aims to reduce the impacts of disasters in the country by adapting pro-active measures as well as cluster roach strategies among government agencies mandated to act upon disaster-related issues. The creation of the Disaster Risk Reduction Management (DRRM) Councils or networks at the national and local levels was aimed to guarantee the coordination and collaboration among government and non-government agencies in the implementation of programmes and policies stipulated in the disaster management plans of the country in order to provide an empowered and resilient Filipino communities. However, since the implementation of the law in 2010, the impacts of typhoons in the country are still enormous. Typhoons Washi in 2011, Bopha in 2012 and Haiyan in 2014 caused severe damage and unimaginable casualties. With this, it can be inferred that the implementation of the policy is poor, and the performance of the Philippine government is ineffective. This article explores the country’s disaster management governance structure and its implications for the performance of the disaster management network of Region 10 Philippines.

Jatmiko and Tandiarrang (2014), in their study of the Indonesian Maritime Agency, found that the existing structure of the agency does not support better communication among agencies which are crucial towards the agency’s performance. Meanwhile, Seng (2013) argued that the polycentric structure of Indonesian disaster management is ideal in responding to the cases of tsunamis in the country; however, it is not suitable to the norms of Indonesian political community. Moreover, Nurmandi et al. (2015) studied the different disasters in Indonesia and concluded that different governance structures are formed in each of the disasters they studied.

Thus, this article substantiates the existing works on the structural analysis of disaster governance networks and addresses the gap in the literature by utilising the mixed-method approach to social network analysis (SNA) in examining how the Philippine disaster management structure affects the disaster governance in the region. Disaster network characteristics and network centrality are particularly generated to analyse the network structure and its implications for disaster governance.

This study uses the mixed-method approach to SNA in evaluating the network structure of Philippine disaster management and its implications for disaster governance. Social network analysis that provided the overview of the network’s topology in terms of its density, diameter and average distance metrics, and of the network’s centrality in terms of degree, betweenness and closeness centrality metrics, determined which organisation or agency held the central role within a network.

A survey was conducted among the target 56 identified disaster response-related agencies and organisations from the disaster management networks of the most susceptible areas in the Philippines – Cities of Cagayan de Oro and Iligan, Province of Misamis Oriental and the overall Region 10 disaster response network, aimed at measuring the existing relationships among member agencies. Forty-four agencies and organisations were able to participate in the survey: Region 10 – 7, Misamis Oriental – 11, Cagayan de Oro City – 14 and Iligan City – 11. Also, key informant interviews were conducted among the following sector: (1) representatives of the lead agencies (see Figure 1), (2) non-government organisations (Philippine Red Cross, Habitat Foundation, Church and Touch Foundation Incorporated) and (3) 10 survivors, on the basis of their involvement in the disaster management operations of the region.

The network’s topology and centrality measures are the main foundation for understanding the network and governance structure of Philippine disaster management. NodeXL software was used to process and analyse the data. Primarily, the topology of the network – density, diameter and average distance – was generated. The network’s density describes the portion of the potential connections with a network that are actual connections. A ‘potential connection’ is a connection that could potentially exist between two ‘nodes’ – regardless of whether or not it actually does. By contrast, an ‘actual connection’ is one that actually exists. To assess the density of the network, the following formula is applied: Total Possible Edges: # Nodes* (# Nodes-1)/2; Density: Actual Edges/Possible Edges. The result of the formula determines if the generated density of the network is considered low or high. Moreover, network diameter is the shortest distance between the two most distant nodes in the network. In other words, once the shortest path length from every node to all other nodes is calculated, the diameter is the longest of all the calculated path lengths. Meanwhile, average distance refers to the shortest path between nodes.

Centrality analysis gives a rough indication of the social power of a node based on how well they ‘connect’ to the network. A highly centralised network is dominated by one person who controls information flow. A less centralised network has no single point of failure. People can still pass on information even if some communication channels are blocked. The centrality of an entity is analysed using the network’s degree, betweenness and closeness measures. Degree centrality measures how connected an entity is by counting the number of direct links each entity has to others in the network; betweenness centrality measures the number of paths that pass through each entity, whereas closeness centrality measured the proximity of an entity to the other entities in the social network.

This study measured the characteristics of the disaster management networks in terms of high or low density, diameter and average distance. Also, the influence of each agency in their respective disaster management networks was assessed based on the centrality measures – degree, betweenness and closeness. Nodes, as referred to in this study, pertain to the disaster response-related agencies and organisations. Specifically, in Table 1, ‘node count’ refers to the actual relationships in the network, while ‘edges’ refer to the agencies and organisation mentioned by each node. Meanwhile, the term ‘lead agency’ or ‘agencies’ refers to the authorised and mandated agencies to lead the disaster-related operations in the country as stipulated in the Philippine law on disaster management or Republic Act 10121.

The data showed interesting findings among the characteristics of the governance networks in the LGUs and at the regional level. Table 1 shows that the Iligan City DRRM network has a 0.219 density score, while Cagayan de Oro has 0.108, Misamis Oriental has 0.169 and Region 10 has a 0.05 density score. The value of the density scores suggests that in the LGUs of Misamis Oriental, Cagayan de Oro and Iligan City, there are a number of connections (high-density) among their respective networks. However, the density of the networks of the Region 10 DRRM Council and the overall DRRM network in the region is low, which implies fewer connections among member agencies in the network. This observation is plausible based on the frequency of the actual interaction among members of the network. When interviewed on 06 December 2016, Ms. A. Caneda explained that:

The comparison of the diameter metric scores reveals that the Misamis Oriental DRRM Council has a diameter of 2 nodes with an average distance of 1.76 nodes. On the contrary, Iligan, Cagayan de Oro, Region 10 and the overall network have a diameter of 4, 5, 4 and 5 nodes, with an average distance of 2.003, 2.371, 2.508 and 2.401 nodes, respectively. These data suggest that the local disaster management networks, primarily the Misamis Oriental DRRM network, have more connections than the rest of the disaster networks. Generally, the networks’ high-density scores, lower diameter and lower average distance are products of a smaller number of network members, which suggest easier familiarity between and among agencies. On the contrary, the regional and overall disaster management networks’ low density and higher average distance suggest a less connected relationship between member agencies. However, the characteristics of the local disaster management networks show that there is high density and higher diameter and the average distance is almost the same as the rest of the networks.

Therefore, member agencies in the regional and local disaster management networks are sparsely connected as revealed in the diameter and average distance scores. However, in the local disaster networks, more connections are established among member agencies as implied by the higher density scores. This situation is validated in the statement of the Iligan City Social Welfare and Development focal person, Ms. P. Mantos. When interviewed on 25 November 2016, she stated that:

Furthermore, these findings suggest that there are processes in the network that generate minimal information sharing which causes ineffective coordination and inefficient disaster response operations in Region 10.

The centralisation scores of the four networks revealed that in terms of degree centrality, the Region 10 DRRM Council has the least degree centrality, while the DRRM Council of the LGUs of Misamis Oriental, Cagayan de Oro and Iligan has high degree centrality. The degree centrality score of the overall DRRM network in the Region is 9.091 degree relationship, which is lower considering the number of its member agencies. Hence, the networks with the smaller number of member agencies and organisations are relatively highly centralised. Members who are mandated by the law to take part in the disaster management activities have higher degree centrality scores than those members whose membership is on a voluntary basis. Thus, the disaster management network at the regional level is less centralised, while the local management networks are highly centralised in terms of degree centralisation, particularly the Misamis Oriental disaster management network. The less centralised character of the regional network is confirmed by the statement of the Region 10 DILG focal person, Ms H. Ocena. When interviewed on 11 December 2016, she stated that:

Meanwhile, in terms of betweenness, the overall network and the Region 10 DRRM network have high betweenness centralisation scores with 54.442 and 88.517 degree relationships, respectively (see Table 1), while the betweenness centralisation scores of Iligan, Cagayan and Misamis Oriental – 14.037, 27.231 and 10.385 degree relationships – are relatively low. These data suggest that the regional offices work as a bridge in the entire network in terms of sharing information and resources during disaster-related operations (see Figure 2). Hence, the regional agencies control the flow of information in the network. Moreover, the Region 10 disaster network may have fewer connections, yet these are significant ties which are vital for the network operations as revealed by its betweenness centralisation score (see Table 1). Additionally, the closeness centrality scores of DRRM networks in Region 10 (see Table 2) suggest the proximity of the regional agencies to the other agencies and organisations in the entire regional network. This validates the finding that the regional disaster management network is highly centralised, while the local disaster management networks are less centralised in terms of betweenness. The highly centralised nature of the regional network in terms of betweenness is validated in the statement of the Region 10 DSWD focal person, Ms E. Cardona. When interviewed on 15 November 2016, she mentioned that:

Table 2 shows the list of agencies from each network with the highest scores according to the centrality measures: degree, betweenness and closeness. Table 2 reveal that the dominant agencies in the DRRM networks of Iligan and Cagayan de Oro are not exactly the agencies who belong in the mandated structure of the National Disaster Response Plan (Figure 1). Interestingly, the involvement of the non-government agencies such as the Philippine Red Cross, Habitat Foundation and Group Foundation Incorporated implies that disaster response-related activities in the LGU could be improved and sustained. Hence, such collaboration needs to be strengthened. When interviewed on 06 January 2017, the representative of the Philippine Red Cross in Iligan City, Mr. G. Galucan, shared that ‘a day after Typhoon Washi hit the Region, we immediately mobilised our volunteers and resources to help in the rescue operations’. The Touch Foundation Incorporated focal person recounted their experience too. When interviewed on 10 January 2017, Mr I. Borja narrated that:

The absence of bureaucratic protocols in the operations of the mentioned non-government agencies is one of the factors that enable them to respond faster and effectively. Noteworthy, the characteristics of the disaster management networks described above – highly dense with sparsely connected member agencies and less centralised in terms of degree – imply that the existing structure of the disaster management networks is not suitable in Region 10 as manifested by its ineffective disaster response operations. When interviewed on 13 November 2016, most of the interviewed survivors of Typhoon Washi shared:

The Iligan City former DRRM Officer, Mr A. Bendijo, expounded that bureaucratic protocols did not work in their favour. When interviewed on 28 November 2016, Mr A. Bendijo said that:

Significant findings are observed from the chosen DRRM networks in the Philippines. Primarily, in terms of the topographic characteristics of the networks, the density scores are high among the LGUs of Misamis Oriental, Cagayan de Oro and Iligan City, while there is a low density in the Region 10 DRRM Council. These data suggest that the smaller the number of network members, the higher the connections between and among member agencies and organisations are established. Similarly, in networks with more members, such as Region 10, connections are hardly established because agencies and organisations are divided among clusters.

Networks with low density imply fewer connections among member agencies. Fewer connections further imply that there are fewer cases or opportunities for face-to-face encounters or similar activities that enhance the quality of the relationship between and among agencies, which lead to fewer interdependences and low trust in the network (Ansell & Gash 2008). Kapucu (2005) noted that effective response and recovery operations require collaborations and trust between government agencies at all levels and between the public and non-profit sectors. Providing incentives fosters inter-organisational communication and trust that enables accelerating inter-organisational network coordination in emergency management response operations (Ansell & Gash 2008; Kapucu 2006; Tang & Tang 2014). Building interdependencies among agencies and organisations through interactive processes increases trust, builds social capital and can develop into collaborative culture which can substantially increase the speed of decision-making and can lead to successful collaborations (Ansell & Gash 2008; Emerson, Nabatchi & Balogh 2012; Jung, Mazmanian & Tang 2009; Kapucu, Arslan & Demiroz 2010; Paraskevopoulos 2010; Shaw & Goda 2004; Shimada 2015). Meanwhile, networks with low density and highly centralised nature are effective conditions for network effectiveness (Raab et al. 2013). However, the regional disaster management network has low density and is less centralised in terms of degree, which suggests weak disaster management structure as characterised by low trust, lack of interdependencies and slow-paced decision-making during disaster management operations in Region 10.

Meanwhile, the presence of the cluster-based lead agency in the region is shown in the centralisation scores of the network (see Table 2). Table 2 shows that in the regional disaster network, the agencies with the most number of connections are considered to be the most important agencies in the network, similar to the agencies mandated by the Republic Act 10121 to lead the disaster response operations in the region according to their respective clusters. Thus, in terms of centralisation scores, the entire disaster management network in Region 10 mirrors the mandated structure in the Republic Act 10121. According to the Republic Act 10121, the disaster management network is structured with a lead agency governing the activities in every cluster and is facilitated by the OCD. In terms of Region 10’s network characteristics, there are only 52 connections out of the 152 mentioned agencies. Hence, network density at the regional level is low. This implies that there is weak collaboration in the regional network, which resulted in the minimal information sharing and less effective and less efficient operations. Consequently, trust and interdependencies were low because of the lack of opportunities to have face-to-face encounters among member agencies at the regional level. These findings confirm the finding of Bharosa et al. (2008) that most agencies in collaborative efforts appreciate the advantages of collaboration but only a few are actually willing to collaborate.

Many scholars believe that disaster management networks should be decentralised. However, Kapucu (2006) argued that decision-making should be centralised to provide clear direction for disaster operations, which should be decentralised. Figure 2 suggests that in the regional network, no single agency leads the entire disaster response operations despite the implementation of the cluster approach. This finding is consistent with the observation during the 2014 Typhoon Haiyan response efforts ‘there was a lack of clarity on who was in charge from the national government’ (Enriquez, 2013). This finding suggests weak collaboration in the overall network of disaster management which resulted in minimal information sharing and less effective and less efficient operations. The minimal network information sharing supports the findings of Jatmiko and Tandiarrang (2014) that new structure should be built in order for information sharing in the network to be strengthened. These findings confirm the theory of Provan and Kenis (2008) that as the network members increases, the network should be governed by a lead-organisation in order to effectively coordinate network activities and decisions.

On the contrary, the disaster management networks in LGUs revealed distinctive findings. The LGUs of Misamis Oriental, Cagayan de Oro and Iligan City have high-density scores with relatively higher average distance and diameter. With a smaller number of member agencies, more connections within local disaster management networks are established.

Moreover, Table 2 revealed that the agencies involved in the ‘actual’ local disaster networks are different from the ‘mandated agencies’ according to the Republic Act 10121 (see Figure 1). According to the actual governance structure (see Figure 2), there is no single agency that dominates their respective disaster response networks, except in the case of the Province of Misamis Oriental, where the role of the Provincial DRRM Office is glaring, being the biggest node in the provincial disaster network (followed by the Provincial DSWD). For the Cities of Iligan and Cagayan de Oro, the disaster networks are governed not just by the designated agencies of the LCE but also by non-government organisations such as the Philippine Red Cross, which acted voluntarily. The latter is not part of the mandated agencies but turned out to be influential and able to exercise leadership during the networks’ disaster response.

Furthermore, the tall structure and the lead organisation form of network governance (Provan & Kenis 2008), which is centralised in nature, do not work in the local disaster management networks because of the dominance of non-government agencies. This finding supports the conclusion of Lester and Krejci (2007) that leadership during disaster management is not about who holds the authority to lead and direct the disaster operations but, more importantly, about who exercises actual leadership in times of crisis. This study strengthens the theory of Bryson et al. (2006) that the ambiguity of membership, which lies in the hierarchy of collaboration where members have overlapping partnerships across networks, and complexity in local environments such as lack of implementation on existing (environmental) policies (Almarez, Penaroya & Rubio 2015) alter the structures among collaborative actions. On the contrary, ambiguity and complexity in local environments are simplified when agreements are attained through an inclusive collaborative process in the network, which is often achieved in a flat-structured organisation, instead of hierarchies (Bryson et al. 2015).

This study further corroborates the findings of Seng (2013) that the structure may be ideal, but it does not necessarily imply that it is suitable in the community as factors such as social norms and political culture might get in the way. This finding also confirms the conclusion of Kapucu and Van Wart (2008) that decentralised decision-making in the form of an excessive reliance on centralised authorities could bring more harm than good, particularly if the authorities are not fully committed to addressing needs and resolving the various challenges along the way. In the case of the LGU in the Philippines, the LCE holds the authority and serves as the emergency manager as mandated by the Republic Act 10121. Hence, emergency managers should fully grasp the value of collaboration by capacitating the members of the network and the community (Kapucu, Arslan & Collins 2010).

Therefore, this study suggests that a highly centralised disaster network with a shared governance and a flat structure should be considered to enhance the competence of the local agencies through an inclusive collaborative process in order to attain agreements, foster interdependencies and sustain reliable partnerships in the region’s disaster management networks. The strong presence of non-government agencies (Table 2) suggests that sustainable partnership or collaboration between non-government and government agencies could lead to a more effective disaster management network, thus better disaster response. With definite and sound government structures, civil society organisations can harness their potential in crisis situations which could go beyond rapid damage assessments (Alegado 2014; Paramita 2012). Thus, adopting a highly centralised network with shared governance in structuring the disaster management networks leads to sustainable and effective structures and processes in the disaster management operations.

Generally, the Philippine disaster management networks in Region 10 failed to respond effectively during Typhoon Washi in 2010 because of significant reasons. Primarily, the overall disaster management network in Region 10 has a low density, which means that trust and interdependencies were low because of the lack of opportunities to have face-to-face encounters among member agencies at the regional level. Moreover, decision-making in the regional disaster management network is less centralised, as revealed by the lack of a dominant or lead agency in the entire regional disaster management. This implies that there is weak collaboration in the regional network, which resulted in minimal information sharing and ineffective disaster response.

On the contrary, the disaster management networks in LGUs have high-density scores with relatively lower average distance and diameter because of the smaller number of member agencies. Hence, the smaller the network, the higher the connection. Despite the relative cohesiveness in the local management networks, the presence and dominance of non-government agencies imply lack of capacities in terms of decision-making and resources among the mandated agencies. With this limitations, the local network structures nurtures the existing bureaucratic protocols which make it more difficult for the local government units to operate effectively. These network characteristics reduce the capacities of the local disaster management networks, which leads to weak disaster operations.

Therefore, the tall structure and the lead organisation form of network governance (Provan & Kenis 2008), which is centralised in nature, do not work in the local and regional disaster management networks in Region 10, Philippines, because at the regional level, such structure does not build interdependencies among agencies, while at the local level, disaster response operations are constrained by bureaucratic protocols, which make disaster management networks less effective. Hence, shared governance should be explored. Structurally, a mixture of the forms of network governance – lead organisation and shared governance – should be investigated. At the national and regional levels, where many organisations are part of the network, centralised decision-making is necessary and disaster operations should be decentralised (Kapucu 2005). However, trust and inter-dependency should be cultivated in centralised networks to come up with effective mechanisms during disasters.