OBJECTIVES:

This study aims to investigate some potential spatial, socioeconomic and demographic determinants of municipal efficiency in combating the COVID-19 using Data Envelopment Analysis and Geographic Inforamtion Systems.

METHODS:

We test the null hypothesis for the Separability Condition that latitude, longitude, population density, social isolation index and the proportion of Bolsonaro's votes can influence the distribution of inefficiencies for different COVID-19 municipalities in Pernambuco, Brazil. The inputs for the cost-efficiency model were the municipal costs for primary healthcare and federal transfers. The output was the number of healthy population. The Moran Index was applied for investigating the spatial concentrations of inefficiencies and their relations with the potential determinants.

RESULTS:

About 2.17% municipalities (4 from 184 cities) are input-oriented efficient and benchmark for COVID-19 best-practices: Calumbi, Jaboatâo dos Guararapes, Recife and Toritama. From the inneficient municipalities, 94 score above 0.5 and 86 below 0.5 and above 0.09. The mean potential for improvement is a 46.80% reduction in the costs to attain efficient prospects. For the separability condition, the reported p-values for the potential determinants are latitude: 0,491, longitude: 0,886, latitude and longitude: 0,773, Social Isolation Index: 0,5135, population density: 0,896, and Bolsonaro's votes: 0,297, all failing to reject the null hypothesis of separability (the exogenous determinants affect the distribution of municipal inefficiencies but not the frontier estimation).

CONCLUSIONS:

The efficiency results report that only a few municipalities made a profitable (social acceptable) use of the resources. All the exogenous determinants affect the distribution of inefficiency at 95% confidence. The higher impact towards potential shifts in the efficiency frontier estimation comes from cities' geographical position (latitude) and social isolation, though additional investigations are necessary. The Moran index, despite small, indicates some evidence for a clustered pattern with the most efficient municipalities close to efficient municipalities, and the opposite is true.