Gradient tree boosting and the estimation of production frontiers

In production theory and engineering, a topic of interest is the determination of technical efficiency of firms from the estimation of a technology. By definition, a technology must satisfy a set of micro-economic postulates. Likewise, a valid estimator of a technology should meet the same set of ax...

Descripción completa

Detalles Bibliográficos
Autores: Guillén García, María Dolores, Aparicio Baeza, Juan, Esteve, Miriam
Tipo de recurso: artículo
Fecha de publicación:2023
País:España
Institución:Universidad Miguel Hernández de Elche
Repositorio:REDIUMH. Depósito Digital de la UMH
OAI Identifier:oai:dnet:rediumh_____::cbc6b74664815940935d21e198261376
Acceso en línea:https://hdl.handle.net/11000/39748
Access Level:acceso abierto
Palabra clave:data envelopment analysis
Free Disposal Hull
boosting
technical efficiency
CDU::5 - Ciencias puras y naturales::51 - Matemáticas
CDU::3 - Ciencias sociales::31 - Demografía. Sociología. Estadística::311 - Estadística
CDU::3 - Ciencias sociales::33 - Economía
CDU::0 - Generalidades.::04 - Ciencia y tecnología de los ordenadores. Informática.
Descripción
Sumario:In production theory and engineering, a topic of interest is the determination of technical efficiency of firms from the estimation of a technology. By definition, a technology must satisfy a set of micro-economic postulates. Likewise, a valid estimator of a technology should meet the same set of axioms. In this paper, for the first time, we adapt the Gradient Tree Boosting algorithm with the objective of estimating production technologies, satisfying the required theoretical conditions. The new approach shares similarities with the standard Free Disposal Hull (FDH) methodology, but with the advantage that it avoids the typical problem of overfitting. Finally, the performance of the new approach based on boosting is measured through a computational experience, determining that the technique based upon boosting decreases the mean squared error by more than 35% with respect to FDH.