Network DEA and Big Data with an Application to the Coronavirus Pandemic

Abstract

Network Data Envelopment Analysis (NDEA) has the potential to be usefully combined with Big Data sets. We first discuss the DEA technology coefficient matrix which incorporates certain Big Data characteristics including volume, velocity, and variety. In addition, we review potential problems that can arise in using DEA to estimate producer’s performance relative some true, but unobserved technology, and proposed aggregation methods to reduce the curse of dimensionality. The various form that NDEA models can take, including dynamic effects, spillovers between producers, joint production of desirable and undesirable outputs, and the reallocation of inputs, across time, to optimize production. An example of the use of NDEA is offered for the Covid Pandemic in the US. We find that an optimal reallocation of tests for Covid could have averted 10,800 deaths.