Multilevel DEA Approach in Research Evaluation.pdfVIP

ISSN 1748-7595 (Online) Working Paper Series Title Multi-level DEA Approach in Research Evaluation Wei Meng and Daqun Zhang Institute of Policy and Management, Chinese Academy of Sciences Wenbin Liu Kent Business School Working Paper No.140 May 2007 1 Multi-level DEA Approach in Research Evaluation Wei Meng and Daqun Zhang Institute of Policy and Management, Chinese Academy of Sciences, 100080, Beijing, China Wenbin Liu Kent Business School, University of Kent, CT2 7PE, Canterbury, Kent, UK Abstract: It is well known that the discrimination power of DEA models will be diminishing if too many inputs or outputs are used. It is a dilemma if the decision makers want to select comprehensive indicators to present a relative ly holistic evaluation using DEA. In this work we show that by utilizing hierarchical structures of input-output data DEA can handle quite large numbers of inputs and outputs. We present two approaches in a pilot evaluation of 15 institutes for basic research in Chinese Academy of Sciences using DEA models . Key words : hierarchical structures, discrimination power, DEA, research evaluation Introduction Nowadays, performance evaluation and benchmarking become routine practice in performance management. It has also been well recognised that a single indicator may not be sufficient for performance management , especially for performance evaluation of research institutions, which normally have multi-dimensionalities of