A parallel data envelopment analysis and Malmquist productivity index model of virtual frontier for evaluating scientific and technological innovation efficiency at universities

As an important component of the national innovation system, universities have received widespread attention in recent years regarding their Scientific and Technological Innovation (STI) efficiency. This study considers a new perspective for viewing talent cultivation, scientific research, and social service as parallel subprocesses for universities to perform their functions. We also combine the concept of virtual frontiers to calculate the STI efficiency at universities, avoiding the situation where multiple Decision-making Units (DMUs) cannot distinguish their rankings due to their effectiveness. In addition, we use the virtual frontier method to calculate the Malmquist Productivity Index (MPI), which avoids the inability of DMU to distinguish the Technical Efficiency Change (TEC) due to continuous period effectiveness. Through empirical analysis, we show (1) Although there are significant differences in STI efficiency among universities, it has gradually increased overall, rising from 0.1623 in 2014 to 0.2433 in 2018; (2) The overall MPI of universities is not particularly ideal, with two periods having an MPI less than 1, and only 14 universities having an average MPI exceeding 1 in four periods; (3) The overall average value for TEC is 1.1342, with only one university having a mean value below 1. On the other hand, the overall average for Frontier Shift (FS) is 0.8775, and all universities have FS values below 1. Finally, based on the research results, several practical suggestions are provided.