The COVID-19 outbreak and the municipal administration responses: resource munificence, social vulnerability, and the effectiveness of public actions

Facing the unprecedented situation of the COVID-19 pandemic, public officials at the municipality-level have no clear benchmarks or tested policies. In this situation, decision-making becomes a controversial process. This article provides insights for public agents in the Brazilian municipalities to deal with the initial stages of the COVID-19 pandemic. We analyzed the actions taken by city halls of the 52 Brazilian municipalities at least thirty days since the first confirmed case of COVID-19. We used a fuzzy-set Qualitative Comparative Analysis (fsQCA) to identify the combinations of contextual factors and public actions that reduced COVID-19 transmission during the critical initial stage. The empirical results show three main paths to guide policy-making: (1) a plural collaboration path involving public and private sectors, operating in a fragile health system; (2) a public action path providing aid programs through intense collaboration inside public bureaucracy; and (3) a resource-based path relying on a well-structured health system.


INTRODUCTION
What is the best response from a local government in the early stages of a pandemic? The pandemic of the coronavirus disease 2019 (henceforth  is an unprecedented situation in modern times. The serious threat to public health promotes the debate about actions taken by government officials. However, without consolidated and tested policies, decision-making becomes a controversial process. In this context, specific information or action can be considered either exaggerated, worrying, or insufficient, depending on the perspective of the analyst (Melo & Cabral, Forthcoming). Despite the great number of studies regarding the epidemiological aspect of COVID-19, the evidence of the effectiveness of government acts is still scarce.
Until April 16th, 2020, when we started the data collection process for this study, Brazil had almost 29,000 confirmed cases of COVID-19 and more than 1,700 reported deaths. The first case appeared in the city of São Paulo on February 25th. Since then, 1,152 municipalities have been dealing with at least one confirmed case of COVID-19. It took 51 days for coronavirus spread to more than 20% of the municipalities in a country of continental size. Brazil is large and heterogeneous, and the municipal governments face the COVID-19 outbreak in quite different contexts and possess a diverse set of resources and capabilities. This article is an attempt to identify which public administration actions, at the municipality level, have reduced the spread of COVID-19 more successfully.
To do this, we collected data from 52 Brazilian municipalities where at least thirty days had passed since the first COVID-19 case was confirmed on April 16th. We focused on the first 30-day period because we understand that acts taken during this initial phase would have a lasting impact, given the exponential spread of the disease. In the data collection process, we developed measurements to capture two broad conditions in each municipality, namely: (i) the contextual features of the municipality; and (ii) the governance structures and actions taken by the city halls in the fight against the disease. As one could imagine, there is not a silver-bullet to succeed in the fight against COVID-19. Therefore, it would be naïve to search for a unique solution for public administration efforts. Thus, instead of relying on econometric techniques, we used the fuzzy-set Qualitative Comparative Analysis (fsQCA), which identifies the different possible paths leading to success. In other words, fsQCA enables us to arrive at different combinations of conditions (aka. solutions) that simultaneously lead to a successful outcome, in our case, a reduction in the spread of COVID-19.
The empirical evidence identifies three main paths consistent with the reduction in transmission of COVID-19 during the initial stages of the outbreak. The first path taken to avoid the spread of the virus was developing plural collaboration across public units and involving private sector partners to compensate for the lack of specialized resources. We call this path the plural collaboration path. The second path is the implementation of government aid programs through collaboration within the public sector to overcome the lack of health resources and the social vulnerability of the local population. We call this path the public action path. Finally, the third path is the reliance on the wellstructured public-health system and coordination settings within a single public unit. We call this the resource-based path.

PUBLIC ADMINISTRATION AND THE RESPONSES TO COVID-19 OUTBREAK
Although COVID-19 is a still new disease of little-known parameters, the medical and scientific communities quickly reached consensus on practical measures to reduce transmission and react to the outbreak 1 . Until now, some principles are guiding government officials, including: (i) isolation of confirmed and suspected cases as well as tracing who had contact with a potential virus carrier (Hellewell et al., 2020); (ii) mobilization of human and material health resources, given that the quick proliferation of the disease leads to an overwhelming number of hospitalizations, and the possible subsequent collapse of the health system (Remuzzi & Remuzzi, 2020); and (iii) generalized social distancing or quarantine, to prevent the proliferation of the disease and to alleviate the burden on the health system, i.e. flattening the spread over time (Lazzerini & Putoto, 2020). From the public administration standpoint, these principles are converted into practical actions, under locational specific conditions. Contextual conditions. In the public administration literature, the context interacts with the management practices to impact the performance (O'Toole & Meier, 2015). The availability of resources in the surroundings is called environmental resource munificence. When resources are abundantly found, a larger number of opportunities are in place, and this increases the organization's performance (Andrews, 2009). In the COVID-19 outbreak, more specifically, the availability of health resources is critical not only to identify and monitor the suspected cases but also to care for critically ill patients (Armocida et al., 2020;Ji et al., 2020).
Another relevant contextual condition in the COVID-19 outbreak is the social vulnerability of the local population. Public health problems are frequently more severe in locations where high percentages of citizens live in socially vulnerable conditions (Ahmed, Ahmed, Pissarides, & Stiglitz, 2020). For instance, there is a current debate in England and in the USA investigating if ethnic minorities are dying of COVID-19 in disproportionately higher numbers than white people 2 . Furthermore, the vulnerable population is living in densely populated areas with poor infrastructure, which hinders disease prevention (Snyder, Marlow, Phuphanich, Riley, & Maciel, 2016). Thus, social vulnerability should also affect the results of policies to contain the spread of COVID-19.
Public administration actions. The first governmental responses to the COVID-19 outbreak are restrictions on the movement of the population in order to decrease virus transmission (Chen, Yang, Yang, Wang, & Bärnighausen, 2020). Subsequently, public administrations have to address a secondorder impact of massive social distancing. As mentioned, social vulnerability can play an essential role during the outbreak. However, the informal economy is significant in Brazil, and suddenly poor people have their income flow interrupted by staying at home for a few days. Thus, it can be harder to keep a social distance in the presence of social vulnerability 3 , hindering the prevention of the disease dissemination. During the COVID-19 outbreak, the provision of government aid programs to meet the basic needs of the vulnerable population 4 .
Governance for public administration actions. More recently, the perspective of a unilateral policymaking process is increasingly being replaced by collective decision-making involving public and private sectors (Ansell & Gash, 2007). In this scenario, public administration has to find ways to achieve collaboration across multiple public units through collaboration within the public sector (Cabral & Krane, 2018;Sedgwick, 2017). Notwithstanding, the solutions are not only concentrated within public bureaucracies. Collaboration with the private sector is critical to deal with the complexity of the COVID-19 outbreak (Lazzarini & Musacchio, Forthcoming). Inter-sector collaborations work effectively under a myriad of possible configurations (Lazzarini, Pongeluppe, Ito, Oliveira, & Ovanessoff, Forthcoming). On the one hand, inter-sectorial collaborations are efficient under conditions of resource munificence and a stable environment (Provan & Milward, 1995;Raab, Mannak, & Cambre, 2013). On the other hand, while the COVID-19 outbreak does not create a stable environment, network-based organizations can also respond efficiently to situations of crisis and disasters (Kapucu & Garayev, 2011).

DATA AND METHOD
This is a multiple case study of Brazilian municipalities facing COVID-19. The benefit of a multiple case study analysis relies on two main factors. First, it allows us to find common patterns amid different observations considering them deterministically, not probabilistically. In other words, every single case contributes to improve our understanding of the phenomena (Bansal, Smith, & Vaara, 2018;Kahwati et al., 2016). Second, the validity of multiple case studies relies on the selection of cases that share similar a set of characteristics that are anchored on theoretical constructs. Moreover, the possibility of replication of the analysis using similar occurrences validates the empirical effort (Eisenhardt, 1989;Pratt, Kaplan, & Whittington, 2020).
Given that different alternative successful paths are possible, and the interactive nature of public actions, governance mechanisms, and contextual factors, we used a fsQCA method (Rihoux & Lobe, 2009). The fsQCA applies fuzzy set algebra to construct multiple combinations of conditions consistent with a certain outcome (Crilly, Zollo, & Hansen, 2012;Fiss, 2007). In other words, fsQCA maps the different possible paths leading to a similar outcome.
Case selection process and dataset. We identify all Brazilian municipalities with at least 30 days since the first confirmed COVID-19 case until April 16th, 2020. This cutoff date constructed a dataset of 52 municipalities. As municipalities were at different stages of coping with the outbreak on the cutoff date, only statistics and actions that occurred within the first thirty days in each municipality were considered. Box 1 provides a list of selected municipalities. It is important to note that a thirty-day threshold delimits the initial stages of the outbreak. Below, we describe the outcome and conditions we used in the model specification. considered a frequency threshold of at least one case in the solution, given the small sample (Rhioux & Ragin, 2009). Finally, we reported the intermediate solutions provided by the fsQCA software (Fiss, 2011;Misangyi & Acharya, 2014). Analyzing the sample regarding public action, 26.9% of the municipalities did not offer any government aid to the population. The percentages of those that provided one, two, and three aid programs, respectively, are 51.9%, 19.2%, and 1.9%. Looking at governance mechanisms, in turn, we found that 50% of the city halls used only collaboration within the public sector, 7.7% established collaboration with the private sector only, and 25% created collaboration both within the public sector and with the private sector. Thus, 17.3% of city halls had no collaboration. Table A8 in the appendix presents the descriptive statistics for the contextual factors.

RESULTS
Box 2 presents the results of fuzzy-set qualitative comparative analysis, showing three solutions consistent with preventing COVID-19 transmission in the initial stages of the outbreak. We interpret the results considering that "•" indicates the presence of a condition and "⊗" indicates the absence of a condition. The blank spaces, in fsQCA jargon, mean "don't care" situation; i.e. the condition can be either present or absent for the solution. For instance, solution 1a is characterized by the presence of both collaboration within the public sector and collaboration with the private sector, as well as by the absence of hospital beds. In the first solution, resource munificence in terms of physicians and social vulnerability is a "don't care" condition, which means that these characteristics may be present or not to prevent COVID-19 transmission in this solution. Solution 1b has an absence of all conditions of resource munificence, and the presence of social vulnerability. These contextual factors interact with the presence of government aid and collaboration within the public sector, and collaboration with the private sector is a don't care condition. Finally, solution 2 has both dimensions of resource munificence present in it, whereas all public action and governance conditions are absent.

BOX
Before the discussion, it is worthwhile commenting on the social distancing procedures taken in the municipalities. First of all, despite its importance, social distancing laws were not explored in the model, because this was mainly a decision taken at the state-level rather than at the level of municipality. Second, all municipalities adopted social distancing laws, and it seems reasonable to assume that social distancing is a ubiquitous practice in the sample. However, it is possible to track how many days it took from the first case to the closure of shops and schools in each municipality. In general, comparing the number of days per solution, only municipalities in solution 1b seems to take longer to close the schools than the sample's average behavior. Box A9 in the appendix summarizes the analysis 9 .

DISCUSSION
Insights for municipality-level government officials. The results demonstrate that public action options are limited by contingent conditions in the municipality's environment, and there are three possible paths for policymaking. Figure 1 presents a visual representation of the decision-making process based on the results 10 . This can serve as a reference for public officials coping with the initial stage of COVID-19 outbreaks in their locations.
The first path (solution 1a) is limited by the lack of specialized resources, therefore, to avoid the spread of COVID-19, the local government has to develop plural collaboration across the public and private sectors. An example is a public-private partnership between the city hall of Mauá, in the state of São Paulo, and a private hospital to increase the number of hospital beds available in the public universal healthcare system. This path shows that in the absence of resources, collaborations might be the best strategy for local government. Additionally, the Mauá's city hall created the called "Specialized Coronavirus Combat Center", integrating different departments. Therefore, we call this the plural collaboration path. The second path (solution 1b) is constrained by the lack of available resources and threats from social vulnerability. The best strategy for local government is to design public actions such as the provision of aid and coordination with other public units to attenuate the spread of COVID-19. For example, the city hall of Pinhais, in the state of Paraná, provided financial aid to informal workers of recyclable garbage collection. Therefore, we call this the public action path. Finally, the third path (solution 2) relies on the well-structured health system. In this case, the presence of healthcare resources guarantees a successful fight against the spread of COVID-19. Therefore, we call it the resource-based path.

Resource-based Path
Note: The figure aims to illustrate the results building upon the interpretation of the QCA solutions. Therefore, the objective here is to facilitate the interpretation but not necessarily to be exhaustive. Source: Elaborated by the authors.
Contributions to the current public administration literature. Comparing the solutions 1 (1a + 1b) and 2, there is an interplay between resource munificence and network-based governance. Resource munificence interacts with the lack of network-based organizations, which contradicts a current public administration framework (e.g., Provan & Milward, 1995), but supports findings of collaborative decision-making in emergency management (e.g., Kapucu & Garayev, 2011). Thus, these findings contribute to a still understudied topic of crisis and disaster management in the field of Public Administration. The role of social vulnerability in the explanations for strategies to cope with public health emergencies can provide new insights into this debate.
Limitations. Taking into account the characteristics of the dataset and the methodological choices, the results are not generalizable. Instead of generalization, the results enable the elaboration of a mid-range theory applied to Brazilian municipality-level during the initial stages of the outbreak in the country. Furthermore, solution 1b highlights the interaction between social vulnerability and government aid programs. However, this solution is also characterized by its location in the Southern Brazilian region and the small population size. Thus, it is important to note that arguing that aid programs are a response to social vulnerability is rather fragile, as the solution 1a shows a don't care situation to social vulnerability in the absence of government aid. https://www.opovo.com.br/coronavirus/2020/04/04/prefeitura-de-fortaleza-lanca-programa-para-producao-de-2-5milhoes-de-mascaras-de-tecido.html

Social vulnerability
Based on quantiles for the whole Brazilian municipalities: Outcome variable Governmental aids 0 = no aid program available; 0.33 = one aid program available; .67 = two aid program available; 1 = three aid program available Collaboration within public sector 0 = "fully in" to the presence of this collaboration type and 1 = "fully out" to the absence Collaboration with private sector Note: The calibration procedure, which is the process of moving from a raw data matrix to a recoded matrix. The recoding method consisted in the application of a fuzzy scale. Source: Elaborated by the authors.