Mobilizing institutional capacities to adapt to climate change: local government collaboration networks for risk management in Mexico City

Rapidly urbanizing, unequal cities such as Mexico City face multiple, interconnected climate hazards (e.g., heat waves, droughts, intense rainfall and flooding, landslides) alongside non-climate risks (e.g., earthquakes). Local governments are on the front line of climate adaptation and disaster risk management but often operate under the tragedy of urgency, limited awareness, and constrained financial capacity. Multilevel governance (MLG) and collaborative networks can enhance institutional capacity by strengthening coordination across governmental levels and with society, improving efficiency via shared responsibilities and learning, and avoiding contradictory policies. Mexico’s federal system is highly centralized, shaping local capacities and dependencies on national and subnational governments for resources and policy implementation. This study asks: what collaborative resources do Mexico City’s local governments have to address multiple risks; how are these resources mobilized into actions for specific risks; and how do local governments interact across jurisdictional and organizational scales? Using an Integrated Risk Management (IRM) approach and a network perspective, the paper evaluates how collaborative resources underpin local actions in a context of urgency and scarcity.

Prior work highlights that technocratic, top-down approaches often fail to manage cross-scale, interacting climate and disaster risks. Multilevel governance facilitates both horizontal collaboration (between local governments, civil society, communities) and vertical coordination (with subnational, national, and international authorities), enabling decentralization, power-sharing, and improved policy coherence. Studies in Mexico and Latin America document how political inequality, centralization, and fragmented policy implementation constrain subnational and local institutional capacities for climate action and disaster risk reduction. Institutional capacity—consensus-building, coordination, strategy-setting, and resource mobilization—is critical for effective adaptation. In Mexico, legal mandates assign local governments key responsibilities (e.g., land-use planning, public services), yet strong dependencies on higher levels persist for funding and technical capabilities. Integrated risk management frameworks and social network analysis have been used internationally to understand how collaborative ties influence preparedness, response, and recovery, suggesting that denser, more diverse networks can improve responsiveness while central actors can expedite information and resource flows. This study builds on these insights by empirically mapping Mexico City local governments’ collaborative resources and networks across priority risks.

Design: A mixed-methods approach combined participatory processes with Integrated Risk Management (IRM) analysis and Social Network Analysis (SNA). Data collection included in-person participatory workshops (2019–2020), virtual workshops, and an online survey with officials from the 16 Mexico City local governments (alcaldías). Focus risks were prioritized by participants (earthquakes, flooding, wildfires). Collaborative resource matrix: Existence of collaborative resource categories was captured as binary data and integrated into a matrix to characterize material, human, financial, and especially collaborative resources, including intra-local government collaboration and external ties (other local governments, subnational, national, and non-governmental actors). IRM actions mapping: Actions were coded via thematic analysis (MAXQDA 2022) following the IRM cycle: (i) risk identification; (ii) forecasting and prevention; (iii) mitigation; (iv) preparedness and relief; and (v) recovery and reconstruction. Frequencies were estimated and normalized (min–max) to compare distributions across governments and risks. Social network analysis: Ego-centered networks were constructed for each local government (ego) from the collaborative matrix and IRM mapping, then integrated by priority risk into three networks (earthquakes, flooding, wildfires). Networks included horizontal ties (e.g., inter-local government) and vertical ties (local–subnational–national; non-governmental). Metrics computed with Cytoscape included density (connectedness and potential collaborative responsiveness), degree centralization (dominant nodes and information/resource flow), and cross-boundary exchange (share of ties by sector: local, subnational, national, non-governmental). Analytic focus: We compared distributions of IRM actions across risks and local governments; characterized network size, composition, and centrality; and assessed differences in horizontal/vertical collaboration patterns.

Collaborative resources: Civil Protection departments at both local and Mexico City subnational levels were central organizers and nodes. Prioritized risks in workshops were earthquakes, flooding, wildfires (plus social risks such as protests and pilgrimages). Local governments emphasized collaborative resources for both vertical (governmental) and horizontal (inter-local and non-governmental) coordination. IRM actions: More than half of actions fell into preparedness–relief and forecasting–prevention; fewer actions addressed mitigation, risk identification, and recovery. Preparedness–relief emphasized drills and brigade readiness (notably for earthquakes). Forecasting–prevention included infrastructure/equipment to reduce flooding (e.g., via FAISM) and wildfire/earthquake prevention. For risk identification, 69% (11 of 16) reported relying mainly on the Mexico City Risk Atlas or lacking sufficient data, often acting reactively where disasters had occurred. Recovery actions commonly required additional technical and financial support. Mitigation was mentioned by only about 15% of local governments. Differences across local governments: Although ~61% of actions clustered in preparedness–relief and prevention, distributions varied by risk and locality. Examples include stronger forecasting–prevention for flooding in Cuajimalpa, Iztacalco, Venustiano Carranza; risk identification in Milpa Alta; mitigation in Gustavo A. Madero and Tlalpan; preparedness–relief in Álvaro Obregón and Coyoacán; recovery–reconstruction in Xochimilco; and balanced distributions in Iztapalapa. In-person workshops allowed exploration of multiple risks per locality, whereas virtual formats typically focused on one due to time constraints. Collaboration networks: Earthquake and flooding networks were similar in size and composition; the wildfire network was smaller. Network metrics: Earthquake—60 nodes, 320 edges, density 0.11, degree centralization 0.38. Flooding—48 nodes, 369 edges, density 0.16, degree centralization 0.52. Wildfires—39 nodes, 93 edges, density 0.09, degree centralization 0.40. Cross-boundary exchange showed dominance of local government ties (62% earthquakes; 62% flooding; 66% wildfires), with subnational at 20%, 20%, and 18%; national at 8%, 7%, and 9%; and non-governmental actors at 10%, 11%, and 8%, respectively. Low densities indicate relatively sparse connectedness; most-connected nodes were the local government egos. Earthquake and flooding networks centered on disaster-response agencies at subnational (Risk Management and Civil Protection) and national levels (e.g., National Center for Disaster Prevention). The wildfire network included a broader set of subnational/national agencies (environment, mobility, human rights) and non-governmental actors (research institutions, neighborhood committees), reflecting the need to access diverse resources under financial scarcity. Highly central local governments included both central and peripheral jurisdictions (e.g., Cuauhtémoc, Coyoacán, Iztapalapa, Álvaro Obregón, Tlalpan), while wildfire centralities aligned with jurisdictions containing large conservation areas.

Findings demonstrate that local governments rely heavily on both horizontal collaboration (especially inter-local government ties) and vertical coordination with the Mexico City Government to address multiple, concurrent risks. The predominance of preparedness and prevention indicates a focus on immediate readiness and risk reduction, yet limited mitigation, risk identification, and recovery suggest capacity and resource constraints, as well as dependence on subnational technical tools (e.g., Risk Atlas) and external expertise. Sparse network densities, combined with moderate degree centralization, imply room to strengthen interconnections beyond ego-centered ties and to reduce potential fragmentation. The stronger and more diverse collaboration observed in the wildfire network underscores the necessity of engaging a wider set of sectoral agencies and societal actors when financial resources are scarce and risks span environmental and social domains. Overall, the evidence supports MLG as a useful framework for coordinating across scales, improving resource use, and enabling more inclusive governance; however, representation of non-governmental and private actors remains limited and warrants deliberate inclusion to enhance knowledge generation, risk identification, and recovery planning.

This study elucidates how Mexico City local governments mobilize collaborative resources within a multilevel governance system to manage multiple risks. Collaboration is most prevalent among local governments and with the subnational government, and actions concentrate in preparedness/relief and forecasting/prevention across earthquakes, flooding, and wildfires. Network analysis reveals relatively low densities and ego-centered structures, with critical roles for subnational and national disaster-response agencies; wildfire risk management involves a broader constellation of environmental, mobility, and human rights agencies and community/academic actors. We identify successful elements of MLG but also gaps: underinvestment in mitigation, risk identification, and recovery; reliance on subnational tools and academic partners; and insufficient engagement of non-governmental and private actors. Future research should build networks from the perspectives of subnational and non-governmental stakeholders, expand horizontal collaboration mechanisms, and strengthen local capacities for data generation, recovery planning, and mitigation, especially under financial scarcity.

The network maps are ego-centered and built from local government participants, which biases centrality toward local governments and underrepresents subnational, national, and non-governmental actors. Non-governmental organizations did not participate directly in workshops, limiting insight into their roles and ties. Virtual workshops often explored a single risk due to time constraints, potentially narrowing action diversity. Many local governments lack in-house data for risk identification and rely on the Mexico City Risk Atlas, constraining proactive planning. Budgetary constraints, especially in jurisdictions with conservation areas and indigenous communities, may limit generalizability of action portfolios across all local governments.