Conference Profa. Dra. Gabriela Sicilia
| 27 July, 2016 | ||
| 12:00 pm |
Title: Addressing the endogeneity issue in DEA applications
Speaker: Gabriela Sicilia
Date: 27/07/2016 12:00 h
Location: Sala de Seminarios, Edificio Torretamarit
Abstract:
The presence of the endogeneity is frequently observed in several economic production processes, however, it has received little attention in the frontier literature and it is overlooked when practitioners apply data envelopment analysis (DEA). Recently, Cordero, Santín and Sicilia (2015) concluded that when one input in the production process is highly and positively correlated with the true efficiency level, endogeneity arises and DEA estimates are flawed. In addition, they find that this decline in DEA performance is further driven by the misidentification of the most inefficient DMUs with low levels of the endogenous input. These findings take on greater significance since high positive endogenous scenarios are similar to those that are likely to be found in many production processes. In this context, the estimation of the technical efficiency using DEA models without taking into account the presence of endogeneity leads to inaccurate efficiency estimates where many of the most inefficient DMUs are identified as benchmarks, which will lead to inappropriate performance-based recommendations. Building upon this evidence, in this research we address two key issues: how can we detect the presence of an endogenous input? And, how can we deal with this problem in DEA empirical applications to overcome this problem and improve estimations? First, we provide a simple heuristic procedure which allows practitioners to identify the presence of an endogenous input in an empirical research. Second, we propose the use of an instrumental input DEA (II-DEA) as a potential solution to deal with the endogeneity problem in order to improve DEA estimations. Monte Carlo results confirm that II-DEA approach outperforms standard DEA when an input has a high a positive correlation with the technical efficiency. Finally, we perform an empirical application to illustrate our theoretical findings.
Brief Bio:
Ph.D. in Economics at Complutense University of Madrid (2015). Her main lines of research are the measurement of efficiency and productivity and causal inference applied to the field of education, combining both methodological and applied elements. Her work has led to several publications in scientific international journals such as the European Journal of Operational Research, Scientometrics, Pacific Economic Review, Latin American Economic Review and The Social Science Journal and have been discusssed in more than 20 national and international conferences and workshops. She has also participated in several competitive research projects and is a regular contributor to the European Foundation Society and Education.
15 July 2016
