A parallel network BoD model

This paper uses a parallel Data Envelopment Analysis (DEA) modelling approach in order to construct a Total Public Sector performance index which consists of seven dimensions: Administration, Education, Health, Public Infrastructure, Distribution, Stability and Economic Performance. In order to aggregate the seven sub-indicators, the model uses a single constant input DEA model, which is often referred to as the Benefit-of-the-Doubt (BoD) model. There are three different approaches in modelling a parallel DEA model relative to its inputs. The first approach assumes that that the inputs are dedicated to each subprocess. The second approach assumes that the inputs are shared among the subprocesses which perform similar activities; therefore, they have the same multipliers. The third approach assumes that the inputs are shared among the subprocesses which perform different activities; therefore, they have different multipliers. This paper proves that in the case of the BOD model, the three approaches are observational equivalent. Although, the efficiency scores are the same across the three models, the input multipliers, which are also the same, have a different interpretation depending on the type of inputs.