Network of Networks: Preliminary Lessons from the Antwerp Port Authority on Crisis Management and Network Governance to Deal with the COVID-19 Pandemic.

In this article, the authors describe and illustrate what they call a "network of networks" perspective and map the development of a lead network for the Antwerp Port Authority that governed organizations and networks in the port community before and during the COVID-19 pandemic. They find that setting a collective focus and selective integration are crucial in the creation and reproduction of an effective system to adequately deal with a wicked problem like the COVID-19 pandemic. The findings on crisis management and network governance are used to engage practitioners and public policy planners to revisit the current design and governance of organizational networks within organizational fields that have been hit by the COVID-19 pandemic.

A network is a system of three or more organizations (undertaken either voluntarily or by mandate) that work together to achieve a purpose that none of the participating organizations could achieve independently by themselves (Provan, Fish, and Sydow 2007; Provan and Kenis 2008). They are distinct entities with unique identities that require examination as a whole (Provan, Fish, and Sydow 2007; Provan and Kenis 2008; Raab and Kenis 2009). Despite the prominence of networks in practice, their popularity as a research subject, and their relevance for society (Raab and Kenis 2009), we still tend to study individual organizations to understand the collective behavior of networks.

Studies of networks from multiple disciplines predominantly focus on organizations and their relations (ego networks), potentially neglecting or even misinterpreting the relationship between the details of the network and the larger view of the whole (Bar‐Yam 2004; Provan, Fish, and Sydow 2007). In addition, we sometimes forget that studying networks from a whole network perspective is necessary but not sufficient to understand “such issues as how networks evolve, how they are governed, and, ultimately, how collective outcomes might be generated” (Provan, Fish, and Sydow 2007, 480). Recently, for instance, it has been argued by Nowell, Hano, and Yang (2019, 214) that an external (outside‐in) perspective should accompany our dominant internal focus on networks to explain “the forces that may shape and constrain action in network settings.” Inspired by this so‐called network of networks perspective, this article shows how such a perspective allows for a better grasping of (and hence, dealing with) wicked problems (Cartwright 1987), a type of problem that an organizational field such as a port or a city frequently encounters.

Networks Embedded in Organizational Fields: The Case of COVID‐19

The COVID‐19 pandemic can be understood as a wicked problem because there are no quick fixes and simple solutions to the problem, every attempt to solve the issue is a “one‐shot operation,” and the nature of the problem is not understood until after the formulation of a solution (Conklin 2005). Wicked problems are “defined by a focus, rather than a boundary” (Cartwright 1987, 93). Successfully managing such problems therefore requires a reassessment of how a group of organizations and networks make temporally sense and structure a wicked problem. The COVID‐19 pandemic has directed our attention to a pivotal point in network governance: the connection between complexity and scale (Bar‐Yam 2004). It has led us to acknowledge that an appreciation for the scope and detailed nature of a wicked problem is essential, while simultaneously pairing it with a network solution that matches in scale and complexity (Bar‐Yam 2004).

To understand how to deal with the COVID‐19 pandemic, then, one needs to comprehend the relation between a larger, complex system and the scope and nature of the problem. We call this larger, complex system an organizational field (Kenis and Knoke 2002). In the classics of public administration literature, the relationship between an organization and its environment has been studied from a variety of perspectives, focusing on selection or adaption to institutional pressures and resource dependence (Aldrich and Pfeffer 1976; Oliver 1991). An emphasis on environments is therefore not new. Interestingly, however, network scholars in public administration have only recently intensified their efforts to use concepts of the environment as an explanatory factor in the creation, reproduction, or dissolution of networks (Lee, Rethemeyer, and Park 2018; Nowell, Hano, and Yang 2019; Raab, Mannak, and Cambré 2015).

Building on DiMaggio and Powell's (1983) understanding of an organizational field, Kenis and Knoke (2002, 275) link interorganizational relationships and mechanisms such as tie formation and dissolution to define an “organizational field‐net” as “the configuration of interorganizational relationships among all the organizations that are members of an organizational field.” The key issue here is on which scale and what details we should consider examining intersections of organizations and networks embedded in a certain environment, since environmental dynamics are crucial in our understanding of the creation and reproduction of both system within as well as the larger system as a whole (cf. Mayntz 1993).

However, in order to define and examine such larger, complex systems like organizational fields, we need to understand why organizations and networks come together, cooperate, and consequently create and reproduce such a larger, complex system (Kenis and Knoke 2002; Nowell, Hano, and Yang 2019; Provan, Fish, and Sydow 2007). We therefore propose that instead of focusing on an organizational network as the unit of analysis (Provan, Fish, and Sydow 2007; Provan and Kenis 2008), a shift to a collective of networks that is embedded in an organizational field is instructive (cf. Nowell, Hano, and Yang 2019). This means that our unit of observation shifts from one network as a separate entity with a unique identity (cf. Provan and Kenis 2008; Raab and Kenis 2009) to a network of networks (Nowell, Hano, and Yang 2019).

Network of Networks

Building on Maier's (1998) system of systems approach and using Nowell, Hano, and Yang's (2019) notion of network of networks, we accordingly define a network of networks as an assemblage of networks, which individually may be regarded as subsystems that are operationally and managerial autonomous but are part of a larger, complex organizational field by many types of connections and flows (Maier 1998; Nowell, Hano, and Yang 2019; Provan, Fish, and Sydow 2007).

In this article, we adopt a set‐theoretic approach to network of networks, in line with the long‐standing recommendation by Alexander (2015) Figure 1. In such an approach, a network of networks is “best understood as clusters of interconnected structures and practices” of various networks being distinct entities and having unique identities (Fiss 2007, 1180; Provan and Kenis 2008; Raab, Lemaire, and Provan 2013). This means a clean break from the predominant linear paradigm and instead adopting a systemic view in which we assume that “patterns of attributes will exhibit different features and lead to different outcomes depending on how they are arranged” (Fiss 2007, 1181; Provan, Fish, and Sydow 2007; Provan and Kenis 2008).

Figure 1

Semillatice (a) Versus Tree Structure (b) to Illustrate the Embeddedness of Networks (subsets) in an Organizational Field (Superset)

Source: Alexander (2015). Original text originates from 1965. Figure is adapted from Gabriel and Quillien (2019). By comparing the semilattice structure (a) to a tree structure (b), it becomes clear that “a collection of sets forms a semilattice if and only if, when two overlapping sets belong to the collection, the set of elements common to both also belongs to the collection. Contrary a collection of sets forms a tree if and only if, for any two sets that belong to the collection either one is wholly contained in the other or else they are wholly disjoint” (Alexander 2015, 6–7).

Moreover, we note that often assumptions about the structure and governance of networks are used that are suspect at best for dealing with the complexity that networks bring (Raab, Lemaire, and Provan 2013; Rethemeyer and Hatmaker 2008). Further, most network studies only employ an endogenous perspective on networks, which in some cases is bound to the performance of an individual organization, a network cluster, or a certain organizational domain (e.g., health or social care), despite the fact that networks by nature are multilevel, multidisciplinary, and interdependent (Provan, Fish, and Sydow 2007; Provan and Milward 2001; Raab, Mannak, and Cambré 2015).

In particular, scholars often tend to ignore the specific nature of the problems that networks face in their environments (McChrystal et al. 2015; Raab and Milward 2003). This is an issue because not fully understanding the interdependence of a collection of smaller systems or understanding what the larger, complex system is up against makes dealing with a wicked problem like the COVID‐19 pandemic very difficult.

Scope, Methods, and Network Analysis

As part of a larger applied research project on a collaboration between the fire and emergency services (Antwerp Fire Service, Antwerp Port Authority, police, and municipality, among others) in the Port of Antwerp and Antwerp Management School (Van den Oord et al. 2019), we focus in this article on how the Port Authority of the Port of Antwerp (Belgium) dealt with the COVID‐19 pandemic. In particular, we examine the network structure and the embeddedness of individual actors of both the crisis management team and the leadership team of the Antwerp Port Authority (APA) to describe how this network managed the crisis and governed the port community, which is composed of various organizations and networks, before and during the COVID‐19 pandemic. By providing descriptive evidence concerning the development of the overall network structure and the embeddedness of individual actors before and during the COVID‐19 pandemic, we aim to ground the notion of network of networks and hope to engage practitioners and public administration scholars to rethink current design and governance of organizational networks within their respective organizational fields that have been hit by the COVID‐19 pandemic.

For this article, we narrowed the scope of analysis to two levels to describe the interdependence between crisis management and network governance on the operational and policy level of the APA. As Alter and Hage (1993) suggested, we need to make a minimum distinction between the policy level and the administrative level because coordination of joint efforts tends to transcend organizational hierarchical levels as well as involve multiple different functional units such as divisions or departments. The data allowed us to differentiate between these two levels: the crisis management team (operations, or, to use Alter and Hage's term, administration) and the leadership team (policy).

To understand how the APA attempted to manage the crisis throughout the COVID‐19 pandemic, we conduct a network analysis based on three sources of data. For our primary data source, we draw on the records and minutes of three types of meetings: the crisis management team (CMT) meetings, the nautical partners (NP) meetings, and the leadership team (LT) meetings. The data cover a period of 12 weeks (January 20–April 12, 2020), including 53 meetings mentioning 73 unique actors involved in the port community. Data records are based on 26 CMT meetings with a total estimated duration of 20 hours (66 logbook pages), 16 LT meetings with a total estimated duration of 10 hours (19 logbook pages), and 11 NP meetings with a total estimated duration of 12 hours.

In addition, we consult data from Sciensano, which is the Belgian institute for health responsible for the epidemiological follow‐up of the COVID‐19 epidemic in collaboration with its partners and other health care actors. These data provide insight into the dynamics of the pandemic. The third source of data were coauthors 2, 4, and 5, who managed the pandemic in the Port of Antwerp. The second author attended all CMT meetings and participated in various LT meetings, while the fourth author was present in some of the task force meetings (not examined in this article). By collaborating with these practitioners, we are able to go back and forth to the data during these periods, allowing for interpretation of relationships between the APA and actors in the port community as well as building rich narratives of issues discussed in these meetings.

In table 1, we report descriptive measures of the data on the meetings. For data analysis, meetings were grouped into four phases of the COVID‐19 crisis, with a network structure created for each: (1) pre‐crisis network (January 20–March 1, six weeks), (2) pre‐lockdown network (March 2–March 15, two weeks), (3) lockdown network (March 16–March 29, two weeks), and (4) crisis network (March 30–April 12, two weeks).

Table 1

Types of Data and Descriptive Measures of the Four Networks

	Pre‐Crisis Network	Pre‐Lockdown Network	Lockdown Network	Crisis Network
Period (2020)	January 20–March 01	March 2–March 15	March 16–March 29	March 30–April 12
Weeks	6	2	2	2
Days	42	14	14	14
Meetings (all*)	3	5	24	21
Type of issues~ addressed in meetings	17 total	44 total	195 total	97 total
Actors involved	28	32	57	53
Ties	89	146	482	416
Centralization^	0.25	0.35	0.34	0.54
Density^	0.02	0.03	0.09	0.08

Note: The full table is displayed online.

We had access to data from three types of meetings: crisis management team meetings (N = 25, including three extra), leadership team meetings (N = 15), and nautical partners meeting (N = 10).

Type of issues is based on the crisis management and leadership team meetings (egocentric network perspective).

For the purpose of analysis, values of ties between actors (N = 73) were dichotomized.

The reason we opted for six rather than two weeks in the first period is to illustrate what we observed as a “slow start of the COVID‐19 pandemic that increased exponentially” (Sciensano 2020). This aligns with the meta‐data of the statistical reports of Sciensano, which started issuing data beginning on March 1, 2020, and provided a daily report from March 14 onward.

Four network plots (and one overview plot) are presented to provide descriptive evidence concerning overall network structures and the embeddedness of individual actors in the four phases. For each phase, we present a one‐mode matrix based on the actors’ list in which we weight ties between two actors based on frequency of mentioning in the records of the logbook and/or minutes of the meetings.

Three rounds of coding were executed in an iterative manner in which we went back and forth to the data and codes of various issues and actors involved in each CMT and LT meeting that was reported in the data. In Appendix C in the Supporting Information online, we provide an excerpt of data cleaning and the coding process. We aimed to minimize bias by having the first and second author agree on codes and accordingly discuss the application of codes with the third author to agree on the content of issues and the involvement of actors reported in the various meetings.

Simultaneous with the coding process, an actor list of APA departments was developed, indexed, and pseudonymized (differentiating between operations and policy, n = 18), as well as actors of the port community (n = 55). The coding process was performed in Microsoft Excel. To calculate centralization and density scores reported in table 1, we used Ucinet 6 (Borgatti, Everett, and Freeman 2002). To develop the network plots, we use the node and centrality layout based on degree centrality analysis in the network visualization tool Visone 2.7.3 (http://visone.info/; Brandes and Wagner 2004).

The remainder of the article is organized in three sections. In the first results section, we present the findings on the structure and governance of the network of networks. We display in five figures an overall overview (figure 2), as well as more detailed views for each period (figures 3, 4, 5, 6) to describe the development of the network structure and the embeddedness of individual actors of both the CMT and the LT of the APA during the COVID‐19 pandemic. In the second results section, we elaborate on the findings of how this network of networks managed the crisis before and during the COVID‐19 pandemic. We close with a discussion and conclusion section, in which we present recommendations for future research and practice.

Figure 2

Evolution of Four Network Structures of APA and Port Community during the COVID‐19 Pandemic in the Port of Antwerp

Notes: Centrality layout based on node value (degree std., link strength uniform). Ties with a frequency 1–3 are displayed in light gray. Ties with a frequency 4–6 are displayed in dark gray. Ties with a frequency 7–9 are displayed in black. Ties with a frequency 10–12 are displayed in black with a larger size. Ties with a frequency 13+ are displayed in black with a larger size. Red nodes refer to leadership of APA (dark grey nodes in print). Purple nodes refer to operations of APA (light grey nodes in print). Yellow nodes refer to partners of the Port of Antwerp (white nodes in print).

Figure 3

Pre‐Crisis Network

Notes: Centrality layout based on node value (degree std., link strength uniform). 28 actors with 89 ties in total are displayed. Ties with a frequency 1–3 are displayed. Red nodes refer to leadership of APA (dark grey nodes in print). Purple nodes refer to operations of APA (light grey nodes in print). Yellow nodes refer to partners of POA (white nodes in print).

Figure 4

Pre‐Lockdown Network

Notes: Centrality layout based on node value (degree std., link strength uniform). Thirty‐two actors with 146 ties in total are displayed. Ties with a frequency 1–3 are displayed in light gray. Ties with a frequency 4–6 are displayed in dark gray. Red nodes refer to leadership of PALN (dark grey nodes in print). Purple nodes refer to operations of PALN (light grey nodes in print). Yellow nodes refer to partners of POA (white nodes in print).

Figure 5

Lockdown Network

Notes: Centrality layout based on node value (degree std., link strength uniform). Fifty‐seven actors with 482 ties in total are displayed. Ties with a frequency 1–3 are displayed in light gray. Ties with a frequency 4–6 are displayed in dark gray. Ties with a frequency 7–9 are displayed in black. Ties with a frequency 10–12 are displayed in black with a larger size. Ties with a frequency 13+ are displayed in black with a larger size. Red nodes refer to leadership of APA (dark grey nodes in print). Purple nodes refer to operations of APA (light grey nodes in print). Yellow nodes refer to partners of POA (white nodes in print).

Figure 6

Crisis Network

Notes: Centrality layout based on node value (degree std., link strength uniform). Fifty‐three actors with 416 ties in total are displayed. Ties with a frequency 1–3 are displayed in light gray. Ties with a frequency 4–6 are displayed in dark gray. Ties with a frequency 7–9 are displayed in black. Red nodes refer to leadership of APA (dark grey nodes in print). Purple nodes refer to operations of APA (light grey nodes in print). Yellow nodes refer to partners of POA (white nodes in print).

Because of space limitations, readers can find more detail on the broader research project online. In Appendix A, we provide background on the Port of Antwerp, a description of the APA, and a more detailed account of the two levels of analysis and the involved actors. In Appendix B, we elaborate on each type of brokerage that the lead APA network enacted, illustrating the veracity of this dynamic process governing a network of networks.

Findings on the Structure and Governance of Network of Networks: An Endogenous Perspective on Selective Integration

In figure 2, we map the network structures along the evolution of the COVID‐19 pandemic. In the first six weeks, the network structure was composed of 28 actors dealing with 17 issues in total (see tables 1 and 2). However, in the course of merely four weeks (March 1–29), the number of actors doubled, the number of initial ties multiplied more than five times, and together with its partners, the APA had to deal with 195 issues in just two weeks at its peak. In the last phase, the situation stabilized (March 30–April 4).

Table 2

A Description of Solutions to Problems Differentiated for Scales Highlighting the Interlink between Crisis Management Team and Leadership Team Given Type of Issue

Classifications of Issues (Selective Coding)	Types of Issues (Axial Coding)	APA Operations	APA Directors
Information	Communication	Communication of formal procedure on handling incoming vessels suspected of contamination (January 31) Communication with governments, stakeholders, and service providers how to conduct a case of COVID‐19 contamination (extra CMT meetings on April 6, 7) Guideline for inland navigation sent out for review (April 18t) APA operations organized webinars to ensure national and international partners of the operational readiness and continuation of the Port	First suspect case of personnel possibly infected communicated by intranet of APA (March 12) First communication by CEO around the sixth CMT meeting (March 16) Press conference on wearables (March 24) Contact with government to ensure full operability of Port of Antwerp (March 25) Letter concerning the use of dust masks send out (March 26)
	Confirmation	Confirmation of successful distribution of sanitation products (February 28) Inland navigation plan approved by government (April 8) Confirmation to ship owners that it is safe for their vessels to call at the Port of Antwerp	Unions agreed on measures for working home (March 5)
	Sending or receiving feedback	Feedback received from governments (March 5, 25), stakeholders (March 23, 25), service providers (March 25, April 1) Feedback on situation with the parking of trucks (March 30) Feedback from task force (April 2)	Daily updates from the CMT to the LT Feedback president to board (March 17) Feedback to province (March 27) Feedback from audit (April 2) Feedback on perception of task forces from parties not involved
	Information provision	Information on suspected vessels (March 18, 20, 23, 25, April 6, 10) Information on port transportation, parking, office and equipment, organization (March 23, 24, 25, 26, 27, 30) Information from other ports (March 23, April 1) Internal web page online	Document on temporary employment (March 17) Statement following the press conference of the government on the situation in the Port of Antwerp (continuation of activities) Information from government (March 13) Roundtable with port authorities (March 27) Information from other ports (March 30)
	Inquiry	by FOD VG/GD on March 16 for assistance in the delivery of PPE for the medical sector (turning point in crisis)	From emergency and fire service on border closure (March 20) On financial measures and possibilities (March 20–31)
	Q&A	Start‐up of the corona@portofantwerp.com information address for stakeholders	Multiple questions by mail and telephone received from Port Facility Security Officers (February 10) AP interview with head of communication
	Receiving signals	From Hubei, China (January 22) From suspicion of vessel containment (March 2) From situation on parked truck drivers (March 16) From personnel concerns (March 13, 16, 19, 25) From running out of stocks (March 12), delayed payments (March 24), cargo (March 25) From service providers (March 20) From an incident with incoming vessel (April 14)	From concerns of nautical partners (March 13) From concerns of a port and the task force (March 19) From concerns of customers (March 13) From concerns of service providers (March 17, 19, 24) From concerns of stakeholders (March 17, 18) Identification that one of the key partners might not be fully prepared.
	Sending or receiving updates	Various updates from and to governments, stakeholders, and service providers. Some updates worthy of mention were: Update from cruise ships (March 5) Update from Italian situation (March 12) Update given to social partners (March 12) Update from foreign ports (in heavily affected regions) (March 17) Update from government on safety masks (March 19) Update on cross‐border employees (March 24) Update on biohazardous waste (March 26) Update from Board of Directors (April 3) Update from Media and communicative actions (April 6) Update on cured personnel (April 10)	Update on conference call and telecommunication (March 17) Update from other ports (March 18, April 1) Update on media (March 20) Update on biohazardous waste (March 20) Update on personnel and organization (March 23) Update on task force (March 30)
	Verification	With other port (February 26) With international colleagues (March 12) With stakeholders and service providers (March 16) With customers on inland navigation (March 20)	On necessary measures for continuity and exchange personnel of (March 17)
Decisions	Decision‐making	Various decisions were made. Some worthy of mention were: On travel arrangements, visits, and events (March 2) On in‐house activities and other events (March 12) On start‐up of nautical partners meetings (March 12) On lockdown of operations On safety equipment and personal protection measures (March 17) On safety, medical, and PPE to others (March 20) On reopening of truckers parking (March 30) On lower frequency of CMT meetings (April 2) APA took the lead in developing a procedure was released on how to work with inland vessels. This was released on April 8 and based on earlier decisions and communication, which started as from March 26th.	Various decisions were made. Some mention worthy were: On enhance of business intel (March 17) On implementing extra measure of physical distance in operations (March 17) On involving stakeholders in distribution of disinfectants (March 20) On setup of resilience platform On starting up a team to look into the non‐operational critical processes (March 20) On reduction of frequency of LT meetings (including task force) From March 13, APA took the decision to take the lead in starting up frequent calls with all partners in the nautical chain both on operational as on policy level
Organizing	Division of labor	Of opening contact register for personnel (March 5) Of sanitation of desks and working areas (March 5) Of implementation of HR policy on temporary unemployment and annual leave (March 12 onward) Of implementation of lockdown (March 16)	Of work organization and communication (March 13 onward) Policy on home office and work organization (maximum working from home) Of registration of COVID‐19 related projects (March 17) Of interview arrangements (media) (March 23) Of customers relations for financial help (March 24) Of arrangements with audit committee (March 26)
Planning	Strategic planning	Of defining essential functions in port to allow for travel (March 16) Of post‐COVID‐19 preparation (March 23) Of coordination among governments and emergency and fire services (March 23 onward) Of restart plan (April 6 onward)	To realign organization's long‐term strategy (March 20) To become post‐COVID‐19 ready as a port (March 27)
	Preparation	Of minimum staff in case of lockdown (March 16) Of internal call center Of backup team members for the CMT (March 30)	Preparation of a delegation to the backup Directors to act on behalf of the primary responsible in decision authority (March 18)
Sense‐making	Situational awareness	Business intel (February 26 onward)
	Discussion	On external call center (March 17) On additional risks (i.e., cybercrime) (March 19) On external communication to stakeholders (March 24) On symbolic thank you to all personnel (March 25) On future regulatory advice to Scheldt users (March 27) On authority of inland navigation vessels (April 1) On wearables for physical distancing (April 6t)	On safety masks and equipment (March 18) On parking of trucks/truck drivers (March 23) On the executive order of the government to ensure continuation of the logistic chain (March 25)
	Interpretation	On the perceived need to take measures (February 10) On persons visiting the Port House or the Visitor Centre (March 12)	On the measures taken by the federal government (on March 13) On additional measures taken by the federal government (on March 20)
Solutions	Partial improvements	Public poster in function of first communication (February 27) Campaign of #supporttheport (March 19) Launch of COVID‐19 website (March 19) Launch of external website (March 20) Launch of help line for personnel (March 25)	Poster on physical distancing on request of stakeholders (March 19) Sticker for cross‐border transport (March 23) Support of campaign #supporttheport (April 2)

In those periods, each network structure shows various links between APA actors (purple and red nodes) and external actors (yellow nodes) from which APA derives its legitimacy (Human and Provan 2000). The goal of the APA was to avoid a legitimacy crisis in which it could lose its formal authority over the Port of Antwerp (Human and Provan 2000). Therefore, it aimed to prevent at all costs the closure of the port as a result of the COVID‐19 pandemic. The APA followed what Human and Provan (2000) term a “dual legitimacy‐building strategy,” in which personnel were provided with resources and support to arrange institutions and structures of authority and collaboration such as a CMT on the level of operations and a task force and NP meeting on the policy level. The LT compared with the CMT was more externally focused on building outside‐in legitimacy by solidifying relationships with important stakeholders from the port community.

The APA can be considered successful in managing the pandemic in the sense that the port remained fully operational throughout the four phases on all terminals. In comparison, one U.S. port was reported to have closed down for several hours on March 18 as a result of contamination of personnel (incident report, Port of Houston, March 19, 2020 ).

When comparing the four network structures with each other, we find that after the first six weeks, a core group of actors from the APA assembled. Together these seven actors—(purple and red nodes) displayed in the center of the figures—formed what we term a lead network (cf. Provan and Kenis 2008). Analogous to Provan and Kenis's lead organization—a mode of governance involving a single organization acting as a highly centralized network broker to govern all major network‐level activities and key decisions—a lead network from a network of networks point of view represents a single network composed of multiple functional units from various organizations and networks that differ in (lateral) position, categories of relevant resources, knowledge, or experience, and in proportion to socially valued tangible and intangible assets or resources within the organizational field of the Port of Antwerp (cf. Harrison and Klein 2007).

The comparison of structures with the issues being dealt with before and during the COVID‐19 pandemic (table 1) shows that in the pre‐crisis network (figure 2), the departments and divisions of the APA acted under business as usual. Before the pandemic started, the mode of governance of the APA was best described as brokered governance, which both governs the sustainment of the port community and its activities as well as participates as a broker in major port activities and key decisions (see Appendix A). This corresponds with elements from a lead organization (level of operations) as well as a network administrative organization (policy level) as governance modes (cf. Provan and Kenis 2008).

When we examine the development of the APA evolving from a pre‐lockdown network to a lockdown network to a crisis network, we observe that the governance of the network developed into a lead network that, at its core, was composed of APA actors (cf. Nowell et al. 2018). Over the course of the pandemic, its structure evolved from a state of loosely coupled links that were targeted and appropriate (figures 3, 4) toward a state of tightly coupled links that were stronger and more intense based on the frequency of interactions (figures 5, 6). Note that although the crisis network structure in figure 6 is similar in the number of actors and amount of ties compared with the lockdown network in figure 5, the number of issues to be resolved by the APA in collaboration with others dropped significantly from 195 to 97 in two weeks’ time.

Once the Port of Antwerp entered the phase of lockdown and subsequently crisis, the functional units of the APA took the lead, enacted by multiple brokerage roles allowing them to pool information and resources and work together with port community actors to guarantee the operability of the port as well as ensure that the main sea gateway remained open for shipping. We found incidence for five types of brokerage (Gould and Fernandez 1989; see Appendix B), with the lead APA network selectively integrating various overlapping subgroups both within as well as (in a later stage) between functional units of various organizations and networks.

The strategic orientation in brokering followed by the lead network was to collaborate to achieve the field‐level goal: keeping the port open (cf. Soda, Tortoriello, and Iorio 2018). During the pandemic, this was exemplified by three distinct brokering behaviors: separating, mediating, and joining (stratification of) organizations and networks (Grosser et al. 2019; Gulati, Puranam, and Tushman 2012).

The network analysis also showed that another network was created by the lead APA network to safeguard, monitor, and control the sea gate to the port during the lockdown period (figure 5). In figure 5, this network is difficult to isolate because of the density of the network structure (table 1); however, in figure 6, the network is more evident in the top‐left corner. The inception of the network can be derived from the NP meeting initiated by the APA on March 12. The inception of this network as an institution and structure of authority and collaboration is interesting for several reasons.

The network was highly selective in its member base, representing a limited number of actors responsible for the nautical operations in relation to the Flemish and Dutch ports of the Scheldt estuary, including the port authorities, tugboat companies, and pilots. In line with the shared‐participant network governance mode (Provan and Kenis 2008), this network of a small group of actors aligned around a common purpose: keeping the sea gate to the ports in the Scheldt River open at all costs, despite the fact that these actors have historically been in competition with each other. The priority of keeping each port in the Scheldt estuary open during the pandemic likely explains why they were willing to redistribute operational resources among one another as long as this safeguarded the attainment of not closing down their port.

Although the network was originally incepted as a temporary information diffusion network, its function altered over the course from sharing information, to problem‐solving, to building (inter)national capacity to address future community needs which might arise.

The presidency of the meeting was handed over to the transnational nautical authority over the Scheldt estuary beginning April 1 to be consistent with its formal authority toward external parties (e.g., shipping) and to further enhance consensus and power symmetry between the actors. By stepping down as chair, the lead APA network ensured that competitors remained working together.

Findings on Crisis Management by the Network of Networks: An Exogenous Perspective on Collective Focus

In table 2, we provide an overview of what the APA did in terms of crisis management, differentiating between the levels of operations and policy. When we look at the issues that were addressed in the APA meetings, we see a shift in attention from COVID‐19 as a public health issue toward the effects of the pandemic on the economy and society. Further, the types of issues addressed in the various meetings over the four periods suggest that COVID‐19 as a wicked problem was mostly perceived as a problem of “information provision,” “decision‐making,” and, to a lesser degree, “sense‐making” of the current situation that the APA was in (see table 1 for a summary and table 2 for details).

When we retrospectively examined how the APA managed the crisis in the Port of Antwerp throughout the COVID‐19 pandemic, we found several interesting matters that highlighted the idiosyncrasies of this information problem. Both on the level of operations as well as policy, the APA acquired, distributed, interpreted, and integrated information (Flores et al. 2012). This suggests that COVID‐19 was mostly perceived as an information problem because both a lack and an abundance of information led to not fully understanding the nature of the problem, which made COVID‐19 wicked.

Information was transferred through various means of communication. At some point, the APA even organized webinars to ensure that national and international partners were informed of the operational readiness and continuation of the port. However, in most cases, feedback and updates were exchanged within the APA and with actors from the port community. The APA made sure that information was presented to those who needed to execute particular tasks or coordinated crisis management and communication (Puranam, Alexy, and Reitzig 2014).

Another emerging topic was the operational method, resulting in a clear collective focus on the tasks at hand. This helped the APA get some kind of grip on the crisis situation. Related to this was how the APA developed a collective focus within the port community. Internally, the CMT (operations) reported daily updates to the LT (policy) on various issues. It made sure that it collected the perceptions of parties that were not involved in the task force. The task force was assembled by the APA to have policy‐level meetings with the representative bodies of the main industry and shipping stakeholders, including public actors such as the Federal Police, the Fire Department, the Federal Public Service of Health, and representatives of the municipal, provincial, and regional governments to ensure alignment across logistic chain and the environment in which it acted.

Externally, the APA detected early (warning) signals from the evolving situation in China because of its national and international network of ports. After verifying the signals received, the APA could take informed measures to contain and manage the crisis.

Based on being informed quickly and accurately, the APA was able to take the lead and act proactively. As a response to numerous inquiries about dealing with inland navigation barges within the COVID‐19 context, a procedure was drafted by the APA and shared for consent with the other ports in the Scheldt estuary and with the authorities responsible for inland navigation on March 21 and released on April 8 after final verification with the inland navigation representative bodies. The extensive but delaying consent seeking led to a unified approach toward a highly scattered subgroup, the inland navigation industry, which fully embraced it.

Another example was how the APA prepared and dealt with the lockdown. Belgium went into lockdown beginning on March 18, but on March 16, the APA was already defining the essential functions of the port that needed to remain operational for travel and transportation. The APA's high performance can be (at least) partially attributed to it being principal driven rather than rule driven (cf. Lauder and Marynissen 2018). The APA's guiding principle during the pandemic was that the port should remain open. The data provide various indications that the APA acted in this proactive fashion. For example, on March 13, the APA made the decision to take the lead in starting frequent calls with all partners in the nautical chain, both on the operational and on the policy level. Also, it followed up on the executive order of the government to ensure continuation of the logistic chain (March 25) and realigned its long‐term strategy (March 20), as well as planned how to become a post‐coronavirus port (March 17) with a restart plan (April 6) and a backup team for leadership already in place even before policy was determined (March 18).

This likely explains why we found that from March 23, the crisis was contained, and consequently from April 2, the APA decided to reduce the frequency of meetings. Interestingly, however, when we took into account the actual solutions the APA had devised to solve the pandemic, we found it made a public poster, initiated a digital campaign, launched two websites, and arranged a call center to provide a hotline to help personnel.

Discussion and Recommendations for Future Research

Our findings inform further research on network of networks in public administration in three ways.

First, the notion of network of networks adheres to a multilevel “systems” analysis, which requires both an endogenous and an exogenous view on networks embedded within a larger system (Contractor, Wasserman, and Faust 2006; Maier 1998; Moliterno and Mahony 2011; Nowell, Hano, and Yang 2019; Provan, Fish, and Sydow 2007; Provan and Kenis 2008). The notion of network of networks implies different questions from those asked in the traditional streams of organizational research and network research. It can be crucial in helping public managers and policy planners make their organizational fields more resilient, because it underpins how networks with different configurations of structure and modes of governance have different effects depending on the nature of the problem being faced and the embeddedness of these networks and organizations within an organizational field (Berthod et al. 2017; Raab, Mannak, and Cambré 2015).

Although this article only examined the notion of network of networks from an egocentric point of view (the APA within the Port of Antwerp community), we gained a first glimpse of the scale and complexity that was involved with the COVID‐19 pandemic. Future research could in particular build on and extend this exogenous network of networks perspective, focusing on a collection of multiple networks that in some way are interdependent within an organizational field to explain why and how they might come together to create a larger, more complex system such as a port or city. Based on the preliminary evidence presented here and building on the work of others, we propose two governance mechanisms that can be crucial in these explanations: first, how a network of networks provides and motivates a collective focus by an organizational field on the problem being faced (cf. Kenis and Knoke 2002; Provan and Kenis 2008), and second, how a network of networks installs and uses institutions and structures of authority and collaboration to selectively integrate those networks that need to work together, while others that do not need to work together do not (Nowell, Hano, and Yang 2019; Provan and Kenis 2008; Provan and Lemaire 2012).

The notion of network of networks combined with the network governance perspective of Provan and Kenis (2008) can be helpful in advancing empirical research on the design and governance of network of networks. It centers our attention on the ways in which networks at one level are embedded within and articulated by networks at other levels (Kilduff and Tsai 2003; Tasselli, Kilduff, and Menges 2015). This requires defining the various loci of network governance, for example, differentiating between the use of institutions and the structure of authority and collaboration at various intersections of policy, administration, and task integration (Alter and Hage 1993; Nowell, Hano, and Yang 2019; Rethemeyer and Hatmaker 2008). This not only prompts us to rethink the original dimensions of network governance outlined by Provan and Kenis (2008) from an exogenous perspective, but also suggests adopting a dynamic and set‐theoretic approach toward brokerage (cf. Kwon et al. 2020; Vermeiren, Raeymaeckers, and Beagles 2019).

In addition, previous research on organizational and network design may guide us in how to design institutions and structures on authority and collaboration that can act as structural inter‐ and intra‐faces that a network of networks selectively can employ to achieve a field‐level purpose (Albers, Wohlgezogen, and Zajac 2016; Gulati, Puranam, and Tushman 2012; Moynihan 2009; Kenis and Raab 2020). Specific attention should be given to the influence of various forms of embeddedness and proximity on interorganizational tie formation and dissolution to explain subset and superset relations between various networks understood as smaller systems embedded in the larger, complex system of the organizational field (Mannak et al. 2019; van Zelst, Mannak, and Oerlemans 2017).

Third, the findings provide some preliminary evidence for addressing anticipatory and mitigative actions among a network of high‐reliability organizations, that is, fire and emergency services, police, and municipalities (Weick and Sutcliffe 2011), and networks, that is, the lead APA network and the network of nautical partners (Berthod et al. 2017). The concept of high‐reliability organizations gives direction for anticipating and containing incidents within a single organization and focuses on maintaining a high degree of operational safety that can be achieved by “a free flow of information at all times” (Rochlin 1999, 1554) embedded in “a pattern of heedful interrelations of actions in a social system” (Weick and Roberts 1993, 357). We lack, however, examples of how these types of interrelations have to be designed, supported, or implemented in an organization (Marynissen et al. 2014) or a network (Berthod et al. 2017).

This research helps us understand the response to crisis in a very specific case and context. Nevertheless, several preliminary findings may be generalizable to other organizational fields such as (air)ports, cities, safety regions, health and social care systems, or innovation regions such as the Brainport region. For instance, one important aspect we found was the consistency of communication and the selective integration of organizations and networks with adequate monitoring and control, avoiding imposing strong constraints that limit cooperation or minimize the independence of various subsystems. However, in some contexts such as safety regions, this may be at odds with common practices in crisis management among public organizations that are dominated by a strong command and control approach (Groenendaal and Helsloot 2016; Moynihan 2009).

Moreover, as we are increasingly not only dealing with one specific organization, but with multiple organizational networks that are involved in “taming” a wicked problem, the findings suggests that network managers (brokers) and public policy planners (designers) need to think together about how a network of a collection of organizational networks can create, selectively integrate, and reproduce an effective complex, larger system that offers more adequate functionality and performance to match the scope and detailed nature of a problem that faces an organizational field. Future research needs to determine which configuration of structure and governance of network of networks consistently achieves what field‐level outcomes given the context of an organizational field. When limited diversity is present among various organizational fields, we can start by revisiting the preliminary theorems introduced by Keith Provan and colleagues.

This calls for further investigation of various wicked problems as coevolutionary patterns of interaction between networks and organizations as separate from, yet part of, an environment external to these networks and organizations themselves (Alter 1990; Nowell, Hano, and Yang 2019). The empirical work on fire and emergency services by Berthod et al. (2017), on health care by Nowell, Hano, and Yang (2019), and on public health and infectious disease outbreaks by Raab et al. (2020), De Vries et al. (2019), and Kenis et al. (2019) illustrates both the theoretical relevance and practical value of an exogenous perspective on network of networks.

To stimulate fresh thinking in practice and spur empirical research on network of networks, our viewpoint is as follows: The key to the solution of how to deal with a wicked problem is to structure a system in such a way that provides appropriate incentives for collective focus and selective integration with adequate monitoring and control, but to avoid imposing strong constraints that might limit cooperation or minimize the operational and managerial independence of various subsystems that make up this larger, complex system.

Conclusion and Limitations

In this article, we reported on how the Antwerp Port Authority dealt with the COVID‐19 pandemic by examining the network structure and the embeddedness of individual actors of both the crisis management team and the leadership team. We drew on the records and minutes of three types of meetings: crisis management team meetings, nautical partners meetings, and leadership team meetings. The data covered a period of 12 weeks (January 20–April 12), including 53 meetings mentioning 73 unique actors involved.

The network analysis revealed how the structure of the lead APA network developed during various phases of the crisis. We found various indications of interdependence and emergence between the APA as a lead network in a network of networks within the port community. In addition, the results show how the lead APA network governed organizations and networks in the port community. Practitioners and scholars should be tentative in generalizing these preliminary findings presented here, because the data only allowed us to employ an egocentric network perspective based on the APA lead network. By having provided descriptive evidence concerning the development of the structure, governance, and crisis management by the APA lead network before and during the COVID‐19 pandemic, we hope to engage practitioners and network scholars to rethink current design and governance of organizational networks within organizational fields that have been hit by the COVID‐19 pandemic. It would be very promising for policy and practice to be able in the nearby future to identify what factors of a wicked problem that faces an organizational field determines what combination of structure and governance arrangement we need to employ when, and why.

Appendix S1: Supporting Information

Click here for additional data file.