The Frontier Approach to the Measurement of Productivity and Technical Efficiency Vania Sena Abstract: In 1957, Farrell proposed measuring technical (in)efficiency as the deviation from a frontier isoquant. Since then, various methods have been developed to estimate production frontiers and apply frontier techniques to measure total factor productivity (TFP). This paper presents core techniques for measuring technical efficiency and productivity based on the frontier concept and introduces recent technological advances. Introduction: Recent discussions on productivity and efficiency differentials have advanced the debate on efficiency and TFP measurement. Efficient and productive performance is desirable, and methods must respect economic theory while providing useful information for managers and policymakers. Best practice frontier methods have become popular due to their deep roots in economic theory and ability to decompose productivity change into technical efficiency, technical change, and scale change. The concept of distance from a standard allows operationalization of inefficiency and TFP. The idea of measuring firm performance relative to a best practice frontier dates back to the 1950s. Koopmans (1951) defined technical efficiency as maximizing output for given inputs. Farrell (1957) extended this by measuring inefficiency as deviation from a frontier isoquant. However, the production possibility set is unknown, so research has focused on identifying the frontier. Two methodologies are now available: parametric methods based on econometric estimation of the frontier and non-parametric methods based on linear programming techniques like Data Envelopment Analysis (DEA). During the 1980s, frontier techniques were extended to measure productivity growth. Two main approaches exist: parametric and non-parametric. The parametric approach uses stochastic frontier analysis, while the non-parametric approach is linked to the Malmquist index. Current research in this field focuses on three areas: semiparametric econometrics for stochastic frontiers, statistical properties of DEA estimators, and treatment of undesirable outputs in productivity measurement. The paper introduces Farrell's measure of technical efficiency, parametric methods for frontier estimation, DEA models, and the frontier approach to productivity measurement. It also discusses the Malmquist index and its extension to account for undesirable outputs using directional distance functions. The paper aims to provide an introduction to methods for measuring technical efficiency and TFP based on the best practice frontier, assess their strengths and weaknesses in empirical work, and present methodological advances in this field. It is not an exhaustive survey but includes papers of potential interest to applied economists. The paper is structured into sections on Farrell's measure of technical efficiency, parametric and non-parametric methods for frontier estimation, DEA models, and the measurement of productivity growth and its components.The Frontier Approach to the Measurement of Productivity and Technical Efficiency Vania Sena Abstract: In 1957, Farrell proposed measuring technical (in)efficiency as the deviation from a frontier isoquant. Since then, various methods have been developed to estimate production frontiers and apply frontier techniques to measure total factor productivity (TFP). This paper presents core techniques for measuring technical efficiency and productivity based on the frontier concept and introduces recent technological advances. Introduction: Recent discussions on productivity and efficiency differentials have advanced the debate on efficiency and TFP measurement. Efficient and productive performance is desirable, and methods must respect economic theory while providing useful information for managers and policymakers. Best practice frontier methods have become popular due to their deep roots in economic theory and ability to decompose productivity change into technical efficiency, technical change, and scale change. The concept of distance from a standard allows operationalization of inefficiency and TFP. The idea of measuring firm performance relative to a best practice frontier dates back to the 1950s. Koopmans (1951) defined technical efficiency as maximizing output for given inputs. Farrell (1957) extended this by measuring inefficiency as deviation from a frontier isoquant. However, the production possibility set is unknown, so research has focused on identifying the frontier. Two methodologies are now available: parametric methods based on econometric estimation of the frontier and non-parametric methods based on linear programming techniques like Data Envelopment Analysis (DEA). During the 1980s, frontier techniques were extended to measure productivity growth. Two main approaches exist: parametric and non-parametric. The parametric approach uses stochastic frontier analysis, while the non-parametric approach is linked to the Malmquist index. Current research in this field focuses on three areas: semiparametric econometrics for stochastic frontiers, statistical properties of DEA estimators, and treatment of undesirable outputs in productivity measurement. The paper introduces Farrell's measure of technical efficiency, parametric methods for frontier estimation, DEA models, and the frontier approach to productivity measurement. It also discusses the Malmquist index and its extension to account for undesirable outputs using directional distance functions. The paper aims to provide an introduction to methods for measuring technical efficiency and TFP based on the best practice frontier, assess their strengths and weaknesses in empirical work, and present methodological advances in this field. It is not an exhaustive survey but includes papers of potential interest to applied economists. The paper is structured into sections on Farrell's measure of technical efficiency, parametric and non-parametric methods for frontier estimation, DEA models, and the measurement of productivity growth and its components.