Maturity tool for assessing municipal sustainability: An artificial neural network and data envelopment analysis approach in Brazilian municipalities

Municipalities play a major role in sustainability transition, as they are the source of several negative but also positive environmental impacts (Bithas & Christofakis, 2006) and concentrate the majority of the world’s population, which is directly contributing to and affected by environmental issues (Giles-Corti et al., 2020; Hamstead et al., 2016). Therefore, new studies and approaches to achieve more sustainable cities are indeed important (Krause et al., 2016). Maturity-based tools are widely used in management studies (Aguiar & Jugend, 2022; Pigosso et al., 2013) as a way to assess the current level of the analysed object of study (e.g. operational performance) as well as to support the improvement of it. Recent studies have developed maturity tools for assessing municipalities features such as resilience (Hernantes et al., 2019, Labaka et al., 2019) and smart cities (Warnecke et al., 2019); however, maturity tools to evaluate sustainable development at the municipality level are lacking. Although there is abundant research trying to measure and assess municipalities’ sustainability (Michalina et al., 2021; Tanguay et al., 2010), a maturity-based measurement can contribute by offering a more accessible roadmap for municipalities to achieve sustainable development. This study aims to develop a maturity tool at the municipal level reflecting the four dimensions of sustainable development, i.e. environmental, social, and economical dimensions (Waas et al., 2011), and the institutional dimension (Lima et al., 2022; Michalinaet al., 2021; Moreno-Pires & Fidélis, 2012) and apply it to 530 municipalities in São Paulo State, Brazil. Data was collected for this maturity tool for the four dimensions (e.g. solid waste generated, treated sewage, child malnutrition, life expectancy, unemployment rate, GDP per capita, total revenue collected, and municipal transparency). Firstly, in order to have a technique capable of clustering the municipalities according to their level of sustainable development, our model relied on an artificial neural network (ANN) technique (Debone et al., 2021). This process has been broadly applied for clusterization problems in different fields of study, such as finance, natural sciences, and industry (Kohonen, 2013; Melin et al., 2020). The maturity model was then validated with a performance index developed with Data Envelopment Analysis (DEA) technique. DEA is a non-parametric approach, based on linear programming, that allocates the set of optimal weights in order to maximise the efficiency of the unit of analysis, that is called decision-making units (DMU) (Charnes, Cooper, & Rhodes, 1978; Cook et al., 2014). Results demonstrate that the municipalities can be grouped in three different clusters, i.e. maturity levels, which is helpful to understand the current context of the municipalities in São Paulo state considering sustainable development. The clusters are composed of 140, 263 and 127 municipalities and the three clusters are statistically different among them. Moreover, the approach proposed in this study for developing and validating a maturity model, i.e. ANN-DEA, is one of the main contributions of this research, it can be applied in other areas of knowledge and can also support decision-making and governance for more sustainable municipalities.