A tug of war between centralization and decentralization: the co-evolution of urban governance and water risks in Guwahati, India

Intermediate cities in Asia and Africa house more than 60% of the world’s urban population. More than 54% of the people here face high levels of water stress and are increasingly vulnerable to water-related risks such as floods, limited access to safe drinking water and water pollution. These urban water risks are complex outcomes of climate change, land use changes, governance failures, and social inequities. This study examines the co-evolution of three water risks and governance responses through the case of Guwahati, India. Guwahati, on the Brahmaputra River in the Northeast of India, serves as a case of an intermediate city to demonstrate changes in water risk governance trends. We focus on studying the connected governance responses to changes in three interrelated water risks, i.e., domestic water supply, urban floods, and river pollution. The shifts in the socio-political structures for these risks are assessed by combining historical institutionalism with multi-level perspective theory. The results show that despite the growing diversity of actors and devolution of responsibilities in cities to mitigate water risks, there is an ongoing tug-of-war between centralized and decentralized modes of decision-making. Contrary to the recommendations for decentralized urban governance, the trend is currently tugging towards a centralized decision-making mode. However, when differences arise between the local and national, local actors within their limited frames of action manage to circumnavigate nationally dictated mandates to fit local needs.


Introduction
Over the last fifty years, water risks 5 have dominated the list of most destructive hazards and globally account for more than 74% of disaster-related economic losses (WMO 2021). Within cities, water risk impacts are concentrated due to population densities and experienced differentially due to unequal living conditions (Birkmann et al 2016, Hoekstra et al 2018. Among the factors contributing to urban water risks, the human elements (policies, land-use changes, and social inequities) have a strong bearing on the overall risk experience by modifying where, how long, and who experiences the risks with often those most vulnerable facing the most risk (both water and other compounded risks) (Bai et al 2016, Dodman et al 2022. Intermediate cities in Asia and Africa house more than 60% of the world's urban population, with large numbers (>800 million) living in informal and vulnerable conditions (Satterthwaite 2005, UNDESA 2019). Yet, there is limited insight into how these cities govern and cope with urban water risks (Birkmann et al 2016, Marais andCloete 2017). Through the Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. 5 Risk as conceptualized as 'resulting from the interaction of vulnerability (of the affected system), its exposure over time (to a hazard), as well as the (climate-related) impact and the likelihood of its occurrence' (Hurlbert et al 2019, p. 679). Within cities, the extent of risk experienced, depends on the hazard or disaster event, exposure levels, and vulnerabilities of those exposed (Birkmann 2007, Field et al 2014, Garrick et al 2014 case of Guwahati, this paper expands insights on water risk-related governance processes in intermediate cities of the global South.
Intermediate cities of the global South can face different challenges due to their faster pace of growth, uneven political dynamics, overstretched human resources, and limited financial capacities (Birkmann et al 2016, Westman et al 2019. The definition of an intermediate city varies based on each country's categorization of large versus small cities. Thus, an intermediate city can have anywhere from 50,000 to 5 million population. Therefore intermediate indicates a city that falls somewhere in the middle within a country's spectrum of city sizes (Satterthwaite 2005, Cities Alliance 2019). However, current assumptions on urban governance of water risks, such as institutional structures required for coordinating action, local capacities, resource requirements and insights on private-public partnerships, are drawn predominantly from research on large cities from the global North (Visser and Rogerson 2014, Westman et al 2019). Therefore, the existing assumptions about urban governance may not match the requirements for mitigating and managing water risks in intermediate cities (Beck 2015, Birkmann et al 2016, Véron 2010. This paper maps the governance trends over time by integrating the various levels of socio-political responses to changes in water risks over time for the case of Guwahati (∼1 million population), India (Census of India 2011). The research involves in-depth qualitative explorations of a rich and diverse database spanning 50 years (table 1). As with other intermediate cities, governance of water risk in Guwahati involves multiple actors stretched across various departments (from local municipalities to national-level river basin authorities) with limited resources, thus leading to complicated and disconnected decision-making processes. The presence of the river Brahmaputra and other local rivers compounds water risks in Guwahati. Therefore this paper focuses on three different but interrelated urban-riverine water risks (1) domestic water supply, (2) urban floods, and (3) river pollution. We assess the various configurations of the multi-level participation among diverse actors that are formed and reformed over time to govern these three water risks.
We unpack the changes in physical risks and drivers of institutional and policy change through the following question: How have (de)centralized governance processes co-evolved along with water risks in Guwahati? This question is assessed via two lines of enquiry: • How do institutional structures (multi-level decentralized/centralized) and policy regimes (norms of practice) associated with governing the water risks change over time in Guwahati?
• What are the drivers of these institutional and policy changes? What are the implications of these changes in managing future water risks?
This paper addresses these questions through a socio-politically situated framework that combines a multilevel perspective and historical institutionalism. The conceptual clarity of terms and the combined framework are expanded in section two. We utilize a pragmatic approach in selecting methods to analyze the rich and diverse database drawn from various sources, these details are elaborated on in section three. The results are outlined through a timeline-based narrative in section four. The key outcomes are discussed within section five and concluded via section six. The lessons drawn from this paper contribute to a South-to-South learning (Hollway 2021) into the underlying mechanisms of change in the governance of water risks that can apply to intermediate cities of the global South.

Review of framework and case studied
Governing water risks involves multiple collaborations and nested levels of coordination in response to changing water risks. Hence, understanding the social, environmental, and institutional changes requires a multi-layered approach. We utilize the combined framework of multi-level perspective theory (MLP) (Geels 2005) and Historical Institutionalism (HI) (Thelen 1999, Fioretos et al 2016 as initiated by Roberts and Geels (2019). This framework allows the study of multi-level actors' involvement while accounting for the external (climatic uncertainty) and internal (municipality expansion) influences, along with social and political changes, while identifying conditions and patterns of change (conceptually represented in figure 1) (Roberts and Geels 2019). Such a framework for a city like Guwahati allows the inclusion of various levels of involvement and political processes in the governance of water risks.

Elements of a combined MLP and HI framework included in the research
The multi-level perspective (MLP) theory captures various interactions across different levels and provides a 'big picture' of water risks and their impacts. Though often used to structure socio-technical insights, this theory has extended applications within socio-political contexts (Geels 2005, Nastar 2014, Yasmin et al 2018. Assessing change here involve interactions between landscape (broader events like political party changes), regimes (policy Table 1. Details of data collected from two field visits, web scraping and content mining.
Interviews involved semi-structured questions to allow narration of recent history of water risk governance processes and problems faced (Detailed list Annex1). regimes include widely accepted norms of functioning), and niches (innovative experiments) (Geels 2002). However, MLP has been critiqued as apolitical (Voß and Bornemann 2011), overlooking the horizontal and temporal dimensions that impact social change (Bai et al 2009, Westman et al 2019. Therefore, there is a need for the inclusion of agency and political insight. This gap is where Historical Institutionalism provides the added benefit. Historical Institutionalism (HI) is a research tradition that uses a socio-political lens to gain insights into institutions' origin and progressive transformation (Fioretos et al 2016). HI allows granular attention to the interactions occurring along a horizontal (i.e., within a particular period) and temporal dimension (i.e., across periods) as governance evolves (Fioretos et al 2016, Roberts and Geels 2019, Thelen 1999. However, HI is often narrowed down to a dualism to describe change, but change is more fluid. This dualism is seen in how governance shifts are defined as states of stability or instability. The change in state is based on institutional structure and agents (including endogenous and exogenous agents or actors). Further, change also depends on path dependencies and rule-bound dynamics of agents (i.e., agents' actions are not erratic) (Bell and Feng 2021).
From MLP, we utilize the concepts of 'landscape', 'regime' and 'niche'. We use a conceptual arrangement of path dependency, actor agency, and policy feedback from HI to understand transitions between and within the multiple levels (i.e., local, regional, national, and international). Path dependence refers to future decisions depending on path decisions due to lock-in created by existing infrastructures, including physical, social, and or behavioral (Thelen 1999, Pierson 2000, Fioretos et al 2016. Path dependency also contributes to feedback loops, which can be positive and strengthen a policy regime or negative and weaken or change a policy regime (Weaver 2010, Fioretos et al 2016. Combining HI and MLP allows us to map the patterns and conditions of change (figure 1). Identifying patterns of change allows us to uncover recurring trends, and the condition of change allows us to know when these patterns are perpetuated. While physical changes in water risks can be modelled and estimated, it is challenging to capture or predict complex socio-political dynamics as they are context dependent and based on multiple factors that, at best, can only be partially predicted or modelled (Rayner et al 2005, Cairney 2012). What can be defined, however, are patterns of change that emerge out of the interaction of actors, institutions, and physical risks (Cairney 2012, Berardo andLubell 2019). The patterns of change are identified within the policy regimes when there is sufficient support for a particular change. The factors and actors involved in influencing policy change are diverse. These influences range from politicians, private enterprises, academicians, media, technological advancements, infrastructure and more (Berardo andLubell 2019, Roberts andGeels 2019).
While the MLP and HI framework originates from roots outside urban and water governance literature, it provides insight and structure for this research (Lockwood et al 2017, Roberts andGeels 2019). This combination keeps the research from being apolitical, includes the relevant risks, and allows a context-driven mixed-method study suitable to assess cities in the global South.
Building an integrated understanding, as the MLP and HI framework requires, is an information-intensive process. Below, we outline the data collection process and methods used in this research.

Guwahati and its Rivers, Assam, India
Due to its colonial legacy and a federal democratic structure, India is a rich setting to examine the governmental and non-governmental interactions that have evolved within its intermediate city. Guwahati is the capital of the Northeastern state of Assam and the largest city on the Brahmaputra River (figure 2). Among Indian cities, Guwahati is one of the many fast-growing, intermediate-sized riverine cities Das 2016, Pawe andSaikia 2018). Focusing on Guwahati allows a specific assessment of the governance impacts on water risks for a city on the river within a growing urban context.
Guwahati has experienced multiple hazards, such as monsoon floods and summer water shortages. These hazard experiences allow examination of the co-evolution of water risks and governance responses. The city is situated within the Brahmaputra Valley and interspersed with local rivers (Bahini, Mora Bharalu, Bharalu, Bhasistha, and Lakhmijan) and wetlands of importance such as Dipor Beel, among others (Alam et al 2001, Chakraborty andSingh 2016). This intersection of rivers and wetlands, with urban expansion and restructuring (canalization, encroachments, pollution) of these natural water bodies (figure 5), contributes to water risks such as domestic water supply inadequacies, urban floods, and river pollution. Additional national-level river-related actors are also involved in some aspects of decision-making within the city.

A brief outline of water governance structures in Guwahati
The governance of the three water risks -domestic water supply, urban floods, and river pollution is spread among various departments. This fragmentation in managing water risks is not a new trend and is seen in many cities across India and other countries (Gandy 2008).
Three departments currently handle domestic water supply in Guwahati they include-The Guwahati Municipal Corporation (GMC), the Guwahati Metropolitan Drinking Water and Sewerage Board (GMDWSB or Jal Board) and the Public Health and Engineering Department PHED). Together these institutions supply water to around 40% of the city's population (Bhattacharya andBorah 2014, Hazarika et al 2016). Consequently, many people resort to pumping groundwater or buying water from private vendors, which often costs more than double the flat fee of the public water supply (Bhattacharya and Borah 2014).
The Brahmaputra River is known to flood across the valley in Assam regularly and is expected to increase in intensity under future climate change scenarios (Goswami 2004, Apurv et al 2015. The floods in Guwahati, however, are a result of multiple factors. They include flooding of the Brahmaputra River, high rainfalls locally, plus governance factors such as blockages in the smaller urban rivers due to pollution and land use changes (shrinking wetlands and open areas) (figure 5) (Sarmah and Das 2018). The city's growth has reshaped and infringed on the urban rivers and the surrounding wetlands that used to serve as a sink to store excess floodwaters (Gogoi 2013). To respond to these challenges in a coordinated manner, a consortium of 16 governmental departments began in 2017 coordinated by the Kamrup District Disaster Management Authority (DDMA) (NPb6, GOLa4).
Lastly, there are no large-scale sewage treatment facilities within Guwahati. Most urban wastewater is directly released into the Brahmaputra River and the surrounding wetlands (Bhattacharyya et al 2010, Mahanta et al 2011. As a result, the local rivers of Guwahati are listed among the top ten polluted rivers in India (CPCB 2013). The Assam Pollution Control Board is the primary department responsible for mitigating river pollution.

Methods
Studying co-evolution within the combined framework is challenging. It can run the danger of 'cherry picking' events to match researcher preferences or result in 'random walks' into vaguely related data streams as there are multiple actors, institutions, and changes due to various social, built, and environmental dynamics (Mitleton-Kelly 2003). To avoid 'random walks' and 'cherry picking' among the various historical accounts, we draw on the recommendation of Klein et al (2008) and Hollway (2021). These authors state that the amount of information required to understand the complexity of a system is based on the scale and context within which observations occur. Therefore, in addition to explaining the methods applied, we also define the research boundaries within which the observations are made, the analysis conducted, and the data sources employed.

Method applied: process tracing
Process tracing is a systematic qualitative method to unpack temporal and multi-level institutional insights within a case study (Collier 2011). Through process tracing, a researcher can draw out in-depth and contextspecific descriptions, exploration, and causal connections sequentially to understand patterns of change (Creswell andCreswell 2017, Ulriksen andDadalauri 2016). Further, induction within process tracing is often seated within a specific theory. Therefore, process tracing is used to build on the multi-level perspective and historical institutionalism theories (Roberts and Geels 2019). Trampusch and Palier (2016) outlined ten types of process tracing approaches. Among the types of process tracing available, we employed historical explanation. Historical explanation provides detailed descriptions of internal and external impacts sequenced over time to expand a theory (Trampusch and Palier 2016). Further, this approach also contributes to theory building. However, an essential aspect of reducing ambiguity within such process tracing is outlining the steps involved and maintaining uniformity in analysis over time when drawing inferences (Collier 2011) Our analysis, therefore, follows a sequenced historical exploration in two phases. The first phase chronologically lists different water risk events, related policy interventions, and institutional and political party changes (Smith 2004, Octavianti and. This chronology allows differentiation of specific eras based on dominant and repeated themes such as the eras of political upheavals, multiple policy reforms, economic reforms, etc Further, this chronology also allows listing all the occurrences linked to the water risks. The second phase creates a descriptive report for each era (Collier 2011). Descriptions here aid in diagnosing conditions and patterns that influence governance transitions (Smith 2004, Collier 2011, Siddiki et al 2019. Newspaper content triangulates information and fleshes out missing information (e.g., scandals or complaints) (Leetaru 2013, Barbehön and Münch 2016). These descriptions allow detailed mapping of the pressures and the influence points (Roberts and Geels 2019, Yasmin et al 2018).

Research boundaries in applying process tracing
Creating systematization for drawing governance-based inference within process tracing involves building boundaries to which generalizations are applicable, i.e., applicability can extend to a system that shares common characteristics, occurs within similar areas or exhibits resonance in actor behaviors and modes of functioning (Steinberg 2015). Therefore, the water risks in this case study serve as an anchor to delineate a boundary around what is being studied (Mitleton-Kelly 2003). We identify two boundary conditions to examine urban water risks for this research.
The first boundary is spatial and social, i.e., covering the study site and the extent of governance interactions. This boundary goes beyond the conventional spatial scale, as it is not limited to geographical or watershed borders. Therefore, we examine interactions using the water risks in the city as the center point from which governance interactions moving outwards and inwards are assessed (Allen et al 2017, Gopakumar 2011, Dunn et al 2017. Interactions are examined concerning the three water risks of domestic water supply, urban floods, and river pollution. The interactions are examined up to three scales, and they include: primary-within the city, secondary-state and national, and tertiary-international (Cilliers 1998, Cash et al 2006. The second boundary condition is temporal. But how far behind should one go in a temporal examination of a complex system? The study's relevant time frame can be outlined by identifying the critical junctures. Studying events and states both before and after the juncture is a good practice. However, if one studies future trajectories, only the post-juncture period can be examined (Hollway 2021). Therefore, the temporal boundary for this research begins at the major juncture, which for this research is the post-colonial transitions starting from when Guwahati was selected as the capital of Assam. Further, to reduce excessive noise in the research results, we overlook those events that do not affect the decisions or actions linked to the three water risks (Hollway 2021).

Data sources
It is difficult to access longitudinal governance-related data sets in intermediate cities of the global South (Nagendra et al 2018). A patchwork of various data was collected to supplement and triangulate information to offset this data challenge. Historical and contemporary data from 1970-2019 was collected from multiple sources, including archived documents, interviews, newspaper data sets (Assam Tribune 1970-2019), academic research, reports, and journal articles.
More than 95 individuals were interviewed via the snowball sampling method. The initial interviewees were selected from various arenas, including national, state, and local governments, academics, NGOs, private agencies, media, and politicians (current and ex-ministers), to avoid selection bias (Kirchherr and Charles 2018).
The interviews had open-ended questions such as: how are you involved in the governance of water risks, how do you think the water risks have changed, what concerns do you have, have there been changes in priority of action over time, what mitigation measures have employed, is there a connection in water risks because of the city and its rivers, what are the challenges in mitigating water risk, how is the institutional performance, and what is required to improve and better mitigate water risks (Questionnaire included in annexe).
Of the 95 interviewed, ten recalled incidents from 2005-2018, and the remaining shared contemporary narratives from 2014-2019. The interviews were conducted during fieldwork in September-December 2018 and August-October 2019. Between these two visits, many government officials changed positions. Hence some interviews for high-level positions were done anew for the same posts. The research also included these additional interviews as they provided institutional reflections from another perspective.

Results: transitions in water risks and their governance
We identified and structured the results section for data analysis into eras based on dominant and repeated themes that occurred over time (Hosagrahar et al 2014). These eras include: The narrative in each era includes three parts to maintain consistency. They include: i. Key actors involved.
ii. Policy development iii. Pressures and impacts

Political reforms era 1970-1980
This period reflects the challenges in establishing governance structures and processes as a new city emerges. It includes regional agitations and civil unrest that were demonstrated in Guwahati. The norm or regime of this era was instability and delay. The niche included ideas drawn from experiments in other large cities (e.g., Delhi, Mumbai).

Key actors
The Government of Assam (GoA) was the leading actor in this phase. The national government, and international aid agencies, such as the Ford Foundation and the United National Environment Program, also shaped urban growth. The media played a role in bringing citizens' concerns to the forefront. A priority concern of policymakers was developing a drainage (flood control) and sewerage plan for Guwahati (CMPO 1971, Sarma 1988). The GoA initially approached the Ford Foundation for this plan, which recommended the Calcutta Metropolitan Planning Organization (CMPO). The urban master plan drafted by the CMPO for water supply, drainage, and sewerage was estimated to cost 290 million INR (36 million USD$ in 1971) and was considered expensive for its time (Alam et al 2001, CMPO 1971. Water supply infrastructure was an afterthought and added towards the end of this plan. To mitigate pollution concerns the Assam State Pollution Control Board (1975) was constituted as an autonomous statutory organization in Water (Prevention and Control of Pollution) Act 1974 introduced by the Ministry of Environment and Forests, Government of India.

Pressures and impacts
In efforts to decrease congestion from large cities (e.g., Bombay, Calcutta, Madras), the 4th five-year plan of the national government of India included loans to states for urban development programs (Sivaramakrishnan 2011). GoA capitalized on these loans for the urban development of Guwahati (Government of Assam, 1966, fol. TCP-228/66). However, the Assam Movement or Assam Agitation to retain an Assamese majority and culture, began in early 1947 and was widespread by 1979 (Alam et al 2001, Weiner 1983). This instability sparked civil unrest (big protests) in Guwahati and redirected institutional attention from basic service provision (water provision and O&M) to maintaining law and order (Alam et al 2001). This redirection had ripple effects on future governance responses as it provided a reason for limitations in delivering services.
The lack of infrastructure plan implementation (i.e., water supply, drainage, and sewage) was also attributed to other reasons. This included the suspension of GMC due to corruption charges (Assam Tribune, August 26, 1975, p.1). The lack of a coordinated and dedicated body to implement the urban development plan (Alam et al 2001). The perceived costs for implementing the Master plan  were very high. Even though existing water supply storage (43 million liters per day (MLD) was enough to provide the city's needs, due to 'inequities in water supply' (CMPO 1971, p. 62) access was limited to only 10%-20% of the population. Further, a change in the ruling political party in June 1978 led to the rejection of the CPMO master plans and removed GMC's suspension (Assam Tribune, August 23, 1980, p.4, GoA,1979. Four urban flood episodes (1970,1974,1977,1980) impacted policy outcomes (figures 3 and 4). These floods were big enough to draw political attention and budgetary allocations often in their aftermath. Phrases in the news around causes of urban flood included: the encroachment of wetlands and rivers, lack of clear land allocation policies (Alam et al 2001), high rainfall draining into Guwahati from the surrounding hilly regions, swelling of the Brahmaputra river and local rivers creating a backflow of water into the city, lack of urban planning, changes in drainage patterns of rivers (figure 4), and faulty urban designs (Assam Tribune., July 22, 1980 p.1).

Economic reforms era 1981-2000
This era is characterized by the creation of water infrastructure, implementation of new plans, devolvement of responsibilities to Urban Local Bodies (ULB), and an increase in specialized institutions. The norm of this era was strengthening water supply institutions and infrastructure. The involvement of private enterprise was a niche idea gaining traction for water supply within the government.

Key actors
The main actors included the national government, state government and media. Despite the devolvement of powers to ULB, the state played a substantial role in local decision-making. The new specialized institutions led to the creation of new coordinating institutions. The Guwahati Municipal Corporation (local), Public Health and Engineering Department (state) and Town and Country Planning Department (state) were all involved in implementing the water supply schemes.

Policy and political developments
The 74th Constitutional Amendment Act (74thCAA) of 1992 provided institutional, fiscal, and financial reforms to strengthen and give autonomy to ULB. However, most state and national governments still had control over the revenue streams and therefore continued to influence decisions around urban growth (Sivaramakrishnan 2011). This trend was also observed for Guwahati. The 74th CCA provided structural leverage to create the new ULBs for Guwahati. However, these ULBs were not autonomous and eventually regulated by the state government created Guwahati Development Department (GDD) that was set up in 1994 via an order of the Governor (Order No. 46/93/22) to enable coordination between all the ULBs for Guwahati.
Few niche ideas were drawn from larger cities, such as creating separate organizations to administer basic services (e.g., urban water boards and urban development authorities) (figure 3). Ideas from water supply processes in large cities were adopted in Guwahati through the influence of political coalitions comprising parliamentarians, politicians and others. This resulted in the Assam Urban Water Supply and Sewerage Board Act 1985 (AUWSSB) and the Guwahati Municipal Development Authority Act 1989 (GMDA) introduced as bills in 1985 (Assam Tribune, July 13, 1985, p.1), which were immediately accepted. The institutions were established in 1988 and 1992 to execute water supply and urban Master plans accordingly. Further, to address the ambiguity in land use and ownership, Assam Land Policy (1989) was drafted. This policy is relevant to water risks as it had a loophole that enabled land purchase from squatters, thus legalizing encroachment into wetlands and eco-sensitive hills (Alam et al 2001, p. 41). The previous CMPO plan was scrapped, and a new master plan (1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001) was created to resolve flooding in Guwahati. The previous infrastructure constructed under CPMO plans were left incomplete.

Pressures and impacts
India was a signatory to the UN International Drinking Water Supply and Sanitation Decade. As a signatory, investment support for water supply and sanitation in the urban region more than doubled (Planning Commission of India 1980). The scheme's national five-year plans targeted urban development ranging from provision and services for 'the urban poor' to specific allocations for the growth of small and medium towns (Planning Commission of India 1980, Sivaramakrishnan 2011). These two pressures, and a change in the structure of grants provided to the Northeast Region (90% national government and 10% locally raised), resulted in more than INR 200 million ($1 = INR42 in 1998) allocated towards the construction of water supply schemes for Guwahati. This investment sought to address Guwahati's water demand gap (from 79MLD to 165MLD) (Alam et al 2001, p. 65). The new urban development plan (1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001) also addressed water supply and flooding concerns.
The increase in groundwater usage started during this period with the proliferation of water pumps and bore wells . Before this period, the main drinking water sources were the Brahmaputra River, shallow dug wells, and ponds (Alam et al 2001, CMPO 1971. The sale of domestic water via water tankers and groundwater-based small private supply schemes grew around this period (GOLa32). While filling the water demand gap, this diversification in water supply sources brought out inequalities, as access was limited to those who could afford the cost (Alam et al 2001).
Civil unrest peaked (violent protest) again in the early 1980s, disrupting essential services such as electricity and water provision, water tankers, and reduced operation and maintenance to mitigate floods within Guwahati (Ranjan 2019). This unrest was reduced by 1985 through the Assam Accord, which also, resulted in a government-sponsored technical university, the Indian Institute of Technology-Guwahati (IIT-G), in 1994 (Ranjan 2019). IIT-G became an institution for technical advice for local decision-makers on water supply, flooding, and river-related decisions.
The flood-related plans were propelled by more than seven episodes of urban floods (occurring in 1981, 1984,1985,1988,1991,1998, and 2000). But the flood mitigation plans were on an ad-hoc basis, led by three departments-flood control department, Kamrup district administration and the G-me). On average, after each flood 7.8 million INR (1$ = 42 INR in 1998) was spent on flood damage repair (Alam et al 2001, p. 53). There were barely any plans or news reports on mitigating or reporting river pollution.

Urban reforms and modernization era 2001-2014
Constitutional provisions and interventions in urban water policy by the national government increased in 2000, focusing on medium and small cities (Sivaramakrishnan 2011). Guwahati witnessed an influx of water supply schemes. There was ease in political communication and the flow of funds as the national and state governments belonged to the same political party (Congress). This era included alignment with an overall neoliberal wave and the rise of private-public partnerships (Gopakumar 2012, Sivaramakrishnan 2013. Private companies were mandated to bid competitively for large infrastructure projects in Guwahati (figure 4).

Key actors
Centralized large water supply plans were funded and supported via multilateral and bilateral agreements. Arrangements included a mix of national government grants and international aid agencies such as the Japanese International Corporation Agency (JICA), Asian Development Bank (ADB), and the World Bank (GOLb18, EPb2). With the impetus of the Disaster Management Act 2005, UNICEF played a key role in developing disaster mitigation interventions, creating a quasi-governmental Assam State Disaster Management Authority (ASDMA) by 2010 (NPb6, GOSa4). An active Kamrup District administration also saw the creation and implementation of plans for flood mitigation in Guwahati (GOSa4, GOSb29). For Guwahati to be eligible for external funds (from the national government and external aid agencies), new detailed plans were created for water supply, stormwater drainage, and sewage (Tahal Group (2010), GOLb1). These infrastructures were included as efforts 'to improve the quality of life of its citizens' (ibid.). The new urban master plan (2009-2025) integrated these water supply plans, ensuring that the execution of these plans was monitored by GMDA (GMDA 2009).

Policy and political developments
New institutional plans followed the creation of infrastructure plans to enable effective implementation of infrastructure plans. The JNNURM mission guidelines propelled a push for the creation of a separate water board, the perceived success of the Delhi Jal Board, ongoing water supply institutional challenges, a local political coalition and reports from funding agencies favoring a separate water board (JICA, NJS Consultants Co Ltd, and EXEIDEA Ltd 2008) (Legislative Assembly 2008). The Guwahati Metropolitan Drinking Water and Sewerage Board (Jal Board) was created in 2009 (overtaking the AUWSSB), under the purview of the Guwahati Development Department. The Jal Board was supposed to take over all water supply responsibilities from the different ULBs to reduce coordination inefficiencies. But this move, which is still pending as of [year], was also seen as an inter-departmental and state versus city power grab due to the substantial increase in water infrastructure budgets (GOLb1, GOLb4). ASDMA sought to coordinate the various departments involved in disaster relief and mitigation (GOSa5, GOSA4).

Pressures and impacts
Nationally, a change from schemes to mission mode demonstrated how views on water services shifted from socialist state intervention to economic commodification (Gopakumar 2011, Gopakumar 2012. This change put emphasis on infrastructure creation, devolving power to local bodies, and building the capacity of local institutions to recover costs . This period also saw a continuation of Assam agitations and civil unrest from 1997-2007. This unrest redirected the attention of the local governments from the delivery of basic services to the maintenance of law and order thus, contributing to the neglect of water supply provision and infrastructure creation (GOLb4).
An estimated 101 MLD of public water supply reached only 40% of the city, whereas the demand was estimated at 135 MLD (GMDA 2009). The supply is centered at the core of the city, with reduced access near peripheries and limited account of water lost in transmission (estimated between 40%-50%) (GOLa32) (JICA, NJS Consultants Co Ltd, and EXEIDEA Ltd 2008). The population without supply relied on groundwater, wells, and private water vendors. As in other cities of India, a rise is observed in private water supply through water tankers to fill domestic water supply gaps, bottled water for drinking water (Gandy 2008, Bhattacharya andBorah 2014). While private water vendors offer quick services, the cost is more than double that of public sources, despite being of inferior quality and quantity. ( Private water vendors typically charge between ₹490 and ₹550 for a 2,000-liter tanker, while public water supply changes a monthly flat fee of ₹140 for roughly 21,000 liters per month (GOLa32, EPb4, Epb5)). This difference in price and access exacerbates inequalities and vulnerabilities, specifically during curfews and civil unrest when these water vendors cannot deliver. A news article referred to this challenge.
'K [the] majority of city dwellers are crying hoarse over the failure of the GMC and the public sector agencies to cover them under piped water schemes. Only 37% of the city households are covered by these agencies with their piped water schemes. Newly emerged water pirates have also added to the woes of the commoners.' (Assam Tribune, May 05, 2006).
After flood events in 2001, 2003, 2004, and 2014, infrastructure-oriented management plans were introduced, which included more flood walls along the rivers and dredging rivers (Assam Tribune, May 22, 2010). However, despite these plans, the 2014 flood resulted in more than ten deaths and immense damage, but also instigated a new mode of action to mitigate floods. 'In 2014, 12 people got killed. Anil Nagar, Navin Nagar was underwater for more than 48 h. That is when we really got into addressing floods' (GOLb4). IIT-G proposed a niche idea of Standard Operating Procedure (SOP) that was developed based on academic experiments in flood mitigation. The SOP recommendations included infrastructure, land protection, water management, and interdepartment coordination arrangements. The SOP brought together 16 government departments, academics, and NGOs to work together, leveraging both governance strategies and infrastructure to mitigate floods (GOLb4, and GOLa4).
River pollution continued and in 2009, Bharalu River in Guwahati was declared as one of India's most polluted rivers by the Central Pollution Control Board with BOD at 50 mg l −1 (desired limit is 3mg/l). The subsequent pollution levels at the Brahmaputra River near the city were increasing (CPCB 2013), highlighting river pollution (GOSb22).

Smart city era 2015-2019
This era marks a change in the desired urban growth type coupled with targeted investment provisions. The diversity of actors increased more than fourfold since 1970. Water supply witnessed an increase in partnership modes of operations including private-public, public-public and public-CSO partnerships.

Key actors
Inter-governmental collaboration is seen as more than 16 departments worked together to mitigate floods, along with media, academia, and a few local NGOs. The increase of news media outlets, proliferation of social media, and growth in civil society involvement brought more attention to public grievances. Private actors were involved through formal and informal arrangements. Formal includes policy provision for implementation and construction of large infrastructure projects and informal is through water vendors serving to fill the gaps in demand.

Policy and Political Developments
The mission's focus was broad and could include any (smart) idea that matched the framework of the mission. Strong emphasis was placed on private-public partnerships (Praharaj et al 2018). The 'Atal Mission for Rejuvenation and Urban Transformation' (AMRUT) was also launched in 2015, focusing on water supply and sewerage infrastructure, maintaining open spaces, and transport. AMRUT's ideology, in addition to developing infrastructure, also sought to build capacity and enable urban institutional reform (MoUD 2015).
Guwahati was selected for both the Smart City and the AMRUT mission. Guwahati's smart city plans included reviving urban rivers, wetlands, and developing riverfront infrastructure on the main stem of Brahmaputra (GoA 2016). However, there were reported gaps in the development of plans 'ASDMA was not involved in consultation for 2025 city plans, some presentation on the plans were shared with us but, we are not involved in consultation' (GOLa5 and GOLa6). The AMRUT mission initially focused on sewerage and septage management, transport, and developing green spaces and parks (Guwahati smart city plan, 2016, SAAP 2017). Later, sewage and septage plans were said to be shuffled between the executing bodies -Guwahati Smart City Ltd office and AMRUT mission office (GOLb30, SAAP, 2017).
The 2014  . The SOP included coordination between 16 government departments, media houses, local universities, and civil societies. A specific department, the District Disaster Management Agency (DDMA), under the directive of the District Commissioner (Kamrup Metro) were responsible for coordinating the disaster mitigation effort. The SOP is said to have reduced flood inundation time and provided preemptive action mechanisms (when implemented effectively) (GOLb4, GOLa4, GOLb24). River pollution management was still perceived in terms of infrastructure development and dredging.

Pressures and impacts
The onset of the new government regime in India (Bhartiya Janata Party, BJP) after a decade ruling of the Indian National Congress, (INC) led to a spur of new missions. The Indian government focused on urban growth, as it was stipulated that by 2030, cities would house 40% of India's population and provide 75% of its GDP (Sankhe et al 2010, MoUD 2015. In this era, steps were taken to build partnerships to address urban water risks and services. The partnerships included private-public, public-public, and private-public-Civil Society Organizations (CSO) in various formal and informal arrangements.
The water supply projects that started around 2006-2009 and were supposed to be completed by 2015-2016 are still incomplete. Corruption charges and scams (Baruah 2017, Patowary 2018, extensive monsoon, late payments, and local disruptions are mentioned as reasons for the delay in completion (EPa6). The funding extensions have been conditional based on constructing a sewage treatment plant for the city (EPb2). The water supply gap remains as none of the new projects have been completed. However, the losses in water transmission and the demand for water supply are set to increase. There was also a reported decrease in groundwater due to rising groundwater abstraction and increased groundwater pollution due to inadequate waste disposal (Girija et al 2007, Sarmah andDas 2018). There are no rules against groundwater abstraction on private property in the city, and an ongoing decrease in groundwater levels (Hazarika and Nitivattananon 2016).
Four episodes of high rainfall and river levels resulted in only three flood reports (2016,2017,2018). At the same time, the news reports did not contain information on flood damages but rather brief information on flooding that occurred as river levels increased. However, over time the SOP implementation became diluted. The SOP required regular action, coordination, and willingness to communicate across departments and various organizations, eventually becoming difficult to sustain (GOLa4). Further, increased media attention to flood events created a social and political incentive to showcase popularity via relief measures rather than preventative measures which require routine attention and hence unseen by media (Mb5). A professor working on the technical plan for flood mitigation mentioned 'The technical solutions are fine, but the flood problem of Guwahati, is not a technical problem. It is a social problem. Similarly, it is an economic problem (due to social inequities).' (Aa5).
The smart city project has prioritized the execution of the riverfront development project and a few water ATMs (GOLb20). However, none of the plans for wastewater treatment have been initiated. Wastewater and sewerage that goes into rivers are within the mandate and recent plans of more than six departments and programs, i.e., GMDA, GMC, Smart city limited, AMRUT, Jal Board, Assam Pollution Control Board, and Brahmaputra Board. However, only Assam Pollution Control Board is responsible for river pollution (GOLb22, GOLa22).

Discussion
Guwahati has seen different regime shifts from political instability to a drive for decentralization, an infrastructure focus, and an emphasis on smart city ideals. These shifts occurred as informal water vendors grew in numbers and reach and efforts to consolidate and regulate the new and growing formal institutions expanded under a banner of improving convergence. This evolving nature of institutions and their struggle for agency and action within a growing top-down governance structure as water risks change is seen among many intermediate cities (Birkmann et al 2016, Marais 2016, Roberts and Geels 2019, Zwarteveen et al 2017.

An ongoing tug between centralization and decentralization
New institutions and policies do not neatly replace the old, but rather form an opaque layer enmeshed in responsibility and personnel with the previous entities. These policy layers and the growing diversity of institutions add to the overlap and ambiguity of responsibilities (Zwarteveen et al 2017). We see a constant tug between the local, state and national actors in outlining how Guwahati is governed. When the national government encouraged the decentralisation of institutions within the city, the state created new institutions to oversee and improve the coordination of these decentralised local institutions.
Further, while decentralization is considered a good urban governance practice (Bassi and Kumar 2012), within Guwahati the unfolding externalities of power dynamics, and existing path dependencies, complicate decentralization efforts to address urban water risks. What this means for many growing cities such as Guwahati is that the policies are increasingly decided by national decision-makers based on aspirations and learning from larger cities with a strong global North bias (Bhardwaj and Khosla 2020, Hemani et al 2017, Verrest and Pfeffer 2019. Many of the current regime's practices, such as urban water boards and riverfront developments, centralized water supply systems are adopted based on external experiences rather than local learning (Mohan 1983, Kumar andGeneletti 2015). Further, in federal systems such as India, national governments have evolved to play a substantial role in shaping intermediate riverine cities due to the increasing budgetary allocations and targeted urban growth programs (e.g., Smart Cities program).
This shift towards a central reliance creates a bottleneck of delays, leading to a lack of coordination and fragmentation of actions. Diversifying and devolving decision-making on important aspects such as resource allocation could be one way out of this bottleneck. However, devolving decision-making powers is politically contentious in smaller regions (due to a smaller pool of resources), as seen when efforts to decentralize authority started rising in the late 1990s and then declining in the 2010s in Guwahati.

Circumnavigation of external mandates by local actors
Factors that affect a regime change are not limited to a binary interchange between landscape and niche factors (Voß and Bornemann 2011). The interactions among and within institutions and local actors' interpretation and implementation of policies also contribute to change. Including historical institutionalism within the multilevel perspective framework, highlights these internal political changes (Roberts and Geels 2019).
Most large cities, due to higher internal revenue, bargaining power, and stronger grassroots movements have the capacity to push for local priorities and shape wider urban policy (Gandy 2008, Mundoli et al 2019, Dempsey et al 2017. When urban national programs are drafted, smaller cities such as Guwahati have little influence on mandates created (Sivaramakrishnan 2011. Intermediate cities often have limited resources and rely on state and national government funding. This reliance means an easier acceptance of changes to the governance structures and the adoption of new mandates dictated by the national government or donors (Bhardwaj and Khosla 2020). However, this reliance does not prevent local actors from navigating away from imposed mandates. This circumnavigation of the external mandates by the local actors can be witnessed in the different national and donor programs (JNNURM, AMRUT, SMART city, JICA and ADB water schemes).
Therefore, circumnavigation occurs in how the local actors align locally perceived priorities and needs within national mandate by rephrasing narratives. This rephrasing is witnessed in how the city's water supply infrastructure was negotiated under the JNNURM program as a requirement for social equity. A similar narrative was pitched under Guwahati's banner of sustainability and risk reduction within the AMRUT and Smart City program. Local actors also frequently postpone mandates that are not perceived as a priority, such as the construction of a sewage treatment plant that was planned in the JNNURM plans (2005), which was forwarded to AMRUT (2012) program and then to the Smart City program (2015) and shuffled again to the next urban master plan (2023).
In Guwahati, the state and local actors prioritize infrastructure-based solutions to resolve water risks. This infrastructure prevalence in plans is linked to a positive feedback loop and path dependency based on three reasons. First includes the dominance of technical or engineering-based positions in government departments (Niranjana 2021). The dominance is observed in many large and small cities of the global South (Gandy 2008). Second, financial incentives or fund provisions are often linked to infrastructure creation and finally, the existing infrastructure creates lock-in requiring more infrastructural fixes (Bigger and Millington 2020). This approach can be seen in figure 5, where creating diversions, and restructuring of local rivers as canals or drains has become a flood mitigation solution to ' send flood water faster into the rivers' (GOLa19). Albeit aspects such as improvement of cross-sector coordination and green infrastructural measures (e.g., maintaining wetlands) are slowly gaining some relevance among measures to address water risks. These alternative measures are difficult to coordinate or implement due to limited capacity and local willingness.
Within intermediate cities such as Guwahati, the local government increasingly 'make do' by seizing opportunities, implementing efforts, and renegotiating new risks (feedback risks from interventions) when they occur. However, this 'make-do' attitude also encourages the maintenance of institutional ambiguity, thus making it easy to pass off responsibilities and mistakes (Heinrichs et al 2013, McDuie-Ra 2016.

Implications of increasing diversity of governance-related actors
The actors involved in local governance regimes have increased in number and diversity. Usually, regime shifts can occur every 5-10 years, depending on structure (within the government and private domain), availability to accommodate change, perceived priority, and importantly, a coalition willing to push forward a niche idea (Nastar 2014, Buuren et al 2016. The drivers of change in Guwahati have grown from the national and state governments being the sole decision-makers to now including the influence of academics, international aid agencies, media, private companies, local societies, and political parties. While the number and diversity of institutions have increased to address the complexity, the resulting need for greater convergence and coordination has not been actualized. As mentioned by a local government official, 'K even if you have convergence at the top, percolating it down to the ground level is not easy unless there is a very strong driver at the top' (GOLa15, November 2018). The network of actors governing water risks for intermediate cities function within ad-hoc arrangements due to insufficient human resource and ambiguity of responsibilities.
The increasing actor diversity without effective systemic coordination can create high transaction costs and delays that are too expensive within resource-strapped regions seen in many cities of the global South (Ayres et al 2018). The increasing diversity of actors while increasing transparency can reduce downward accountability (Biddle and Baehler 2019). For example, funds allocated to address water risks are repurposed or adjusted towards other (not risk-related) activities such as building roads, the responsibility for such decisions is ambiguous.
Currently, the highest diversity and number of actors are seen for domestic water supply, followed by urban flood and the least for river pollution. Expenditure preference aligns with this trend. The highest priority is domestic water supply infrastructure, followed by flood mitigation and low interest in sewage treatment options or improving urban river quality. This trend of expenditure has not changed since 1970. This absence of priority for pollution has led to delays in pollution mitigation despite the availability of funds and plans for sewage treatment plants and improving urban river quality since 1970. Such trends and priorities towards risk-related infrastructure investment are not unique to Guwahati. This trend is seen in other cities of the global South (Xu et al 2019) ranging from large cities like Cape Town (Bigger and Millington 2020) to small ones like Surat and Coimbature, India (Acharya 2012, Lele 2018). One way out would be to increase informed citizen action coupled with willing local champions in the government to conserve natural resources such as wetlands and rivers as green infrastructure to mitigate floods and reduce pollution (Jacobs et al 2016).

Conclusion
Our research shows that water risk governance involves a complex interplay of local internal processes, which frequently circumnavigate nationally set mandates. Local priorities associated with water risks rarely include attention to rivers beyond the capacity of stormwater drains. The governance processes have co-evolved based on path dependencies (related to infrastructure and policy), political priorities, and overlapping institutional arrangements, further displaying the growth of mixed messages by overlapping governmental mandates.
Within intermediate cities, understanding the governance trends, diversity and modes of operation involved in addressing water risks can help identify strategic opportunities and areas where intervention will likely  Master Plans: 1971-2001, 1987-2001, 2009-2025 succeed or fail. For Guwahati, a significant aspect includes the ongoing tension between local, state, and national actors as a tug-of-war ensues between centralized and decentralized modes of governance. The ongoing tug is fueled by the growth in the diversity of actors involved in governance. While diversity brings better ideas and transparency, when there are limited systemic avenues for convergence, such diversity creates increased coordination costs and ambiguous responsibilities. This ambiguity can sometimes be seen as an intentional coping mechanism to disperse blame for institutional inefficiencies. Further, the authorities responsible for the river basin are rarely seen to be involved with urban water risk governance, apart from permissions for water allocation, infrastructure construction and data provision related to the river. This low level of involvement coincides with the perception among city and river basin authorities that the urban impact on rivers is negligible.
Policy responses to the challenge of increasing centralization faced in Guwahati include devolving decisionmaking on infrastructure requirements to local authorities, funding (with oversight), and the required human resources for effective execution. Devolution of decision-making also requires a clear mandate to promote strategies such as conserving the natural resources of wetlands and rivers as green infrastructure to mitigate floods and reduce pollution. Reducing the bottleneck of delays and enabling healthy urban and river ecosystems requires local champions to adapt plans to local needs. Such steps would reduce the disconnect between decisions made at a national level versus local requirements. Another policy challenge is the diversity of actors in the governance sphere and the increasing fragmentation of action. This challenge requires dedicated efforts among the top and local actors to build systems that encourage regular convergence and enable clear directives of responsibilities.
However, implementing such policy directives in intermediate cities is difficult due to wider systemic challenges, as these cities are usually on the receiving end rather than informing urban policies. Changes that occur internally in such urban governance processes are influenced by external factors (i.e., national and international development agendas) and driven by political coalitions with limited inclusion of recommendations and local learning taking place. Such external drivers of changes are also observed in other intermediate cities in India. Further, while there is a rhetoric of concern, there is a consistent gap among toplevel decision-makers, city officials and river basin authorities in acknowledging the impact of the city on the river and vice versa.
Changing the impacts of the climate and human-induced urban water risks begins at the local level of governance (Ostrom 2010a). Within this landscape, the nationally driven push towards smart cities and sustainable urban growth requires local actors to prioritize and protect green infrastructure solutions (such as wetlands and rivers) based on local context. Unpacking the changes in local governance processes and their interactions with different levels of governance over time is essential to map overlooked vulnerabilities and opportunities available to respond to future urban water risks and conserve urban rivers.