Purpose: This study aims to measure the impact of public spending on basic education in São Paulo cities on the efficiency of public educational expenditure.
 Method: The calculation of the efficiency is based on the Data Envelopment Analysis (DEA), having as input the public expenditures in education in the years 2014 to 2017 and output the municipal score of Prova Brasil 2017. The measurement of the impact of public expenditures on efficiency is based on the technique of Multiple Linear Regression (MLR) by Ordinary Least Squares (OLS), having as explanatory variables educational expenses classified as wages, purchases, travel, interest and permanent materials, and variable response efficiency index of public expenditure.
 Results: From the statistical models of MLR different regression coefficients were obtained, which estimate the change in the efficiency index of public expenditure when the expenditure is increased or withdrawn. The results expressed that there is a negative correlation between all categories of the expenditure element and the efficiency of public educational spending in the period from 2014 to 2017.
 Contributions of the Study: This study, in a pioneering way, adopts the budget accounting classification of the categories of the element of the expense, thus approaching the study of the efficiency of the closest possible object to which the classification of the expenditure refers.