Measuring dynamic efficiency with variable time lag effects

Abstract

This study is focused on the effect of time lag on a two-stage dynamic network data envelopment analysis (DEA) where intermediate products produced in the first stage both contribute to concurrent production in the second stage and affect operations in subsequent periods. We concluded that existing treatments for estimating a uniform time lag parameter do not reflect various operational strategies deployed by individual decision making units (DMUs) over a period of time. As a remedy, we proposed a two-stage dynamic network DEA model that can consider variable time lag effects, namely multiple carryover schemes, optimized for each DMU in efficiency measurements. Empirical validation of the model is provided based on Korean research and development (R&D) institutes and a focused discussion on the technology licensing process. The comparative analysis shows that the dynamics of intellectual capital stock can be represented well by the proposed model, while the existing approaches carry a risk of understating efficiency measures owing to their restrictive nature. The proposed model is particularly robust to the concentration of inputs and/or outputs by recognizing virtual reallocations of resources over time, which can be used as a solution to the discrepancy between production lead time and interval of evaluation in dynamic systems.