Benchmarking da eficiência financeira no setor de telecomunicações brasileiro

Abstract

The telecommunications sector is one of the strategic pillars of the Brazilian economy, characterized by high capital intensity, rapid technological evolution, and strong regulatory oversight. In this context, the ability of operators to convert economic and financial resources into operating results becomes a determining factor for the sector's competitiveness and sustainability. This dissertation estimates the relative efficiency level of EBITDA in Brazilian telecommunications companies using an econometric frontier approach, analyzing 98 firm-year observations from 12 operators over the period from 2011 to 2024, based on financial statement data standardized in accordance with IFRS norms. The core methodology adopts Stochastic Frontier Analysis (SFA), specified in the Cobb-Douglas functional form and estimated by Maximum Likelihood. The model decomposes the error term into two components (random noise and technical inefficiency), enabling a more precise assessment of firm performance relative to the best-practice frontier, while isolating uncontrollable external factors. The inputs considered were Permanent Assets, Net Revenue, Operating Costs, Commercial Expenses, and Administrative Expenses, with EBITDA as the output variable. Model validity was assessed through formal statistical tests: the Jarque-Bera Test confirmed residual normality, the Breusch-Pagan Test found no evidence of heteroscedasticity, and the Likelihood Ratio Test (LRT) did not identify statistical evidence of systematic technical inefficiency, suggesting relative performance homogeneity across the sector, with the random component accounting for the majority of observed variance. The estimated parameters indicate that approximately 28% of the variability in operating results is attributable to the technical inefficiency component. The estimated efficiency scores, ranging from 0.86 to 0.96, were segmented into three groups using the K-means algorithm, validated by the Kruskal–Wallis Test and Dunn's post-hoc procedure, confirming the existence of statistically distinct performance groups: high, intermediate, and low efficiency. The results indicate that Net Revenue expansion is the primary driver of EBITDA growth, with an estimated elasticity of 1.63, while Operating Cost management exerts a more significant impact than Commercial and Administrative Expenses in determining financial performance. The estimated returns to scale of 1.0014 indicate that the sector operates under conditions close to technological equilibrium, suggesting that additional performance gains depend more on managerial improvements than on simple scale expansion. To strengthen analytical rigor, the SFA results were compared with those obtained through Data Envelopment Analysis (DEA), using the CCR model with output orientation. The Spearman correlation coefficient of 49.85% between the efficiency scores of both methods indicates moderate convergence, superior to that observed in similar studies in the banking sector, reinforcing the relevance of SFA as a benchmarking tool in contexts where exogenous factors and accounting volatility may distort diagnostics based on deterministic methods. As a practical contribution, a Technical-Technological Product (PTT) was developed in the form of a computational script in Python, published on the Code Ocean platform to ensure reproducibility and public access, constituting a practical decision-support instrument for sector managers and analysts. The study contributes to the literature by applying SFA with financial variables to the Brazilian telecommunications sector, demonstrating the methodological viability of the parametric approach relative to traditional non-parametric models, and offering concrete managerial implications for efficiency diagnostics, resource allocation, and competitive benchmarking.