Consequences of Omitting Relevant Inputs on the Quality of the Data Envelopment Analysis Under Different Input Correlation Structures

This paper establishes the consequences of a wrong specification on the quality of the data envelopment analysis. Specifically, the case of omitting a relevant variable in the input oriented problem is analyzed when there are different correlation structures between the inputs. It is established that the correlation matrix gives relevant information about the homogeneity of the decision making units and the intensity of inputs used in the production process. The methodology is based on a series of Monte Carlo simulations and the quality of the data envelopment analysis is measured as the difference between the true efficiency and the efficiency calculated. It is found that omitting relevant inputs causes inconsistency, and this problem is worse when there is a negative correlation structure