Dynamic Data Envelopment Analysis

This thesis deals with the dynamic DEA models, which examine decision-making units during several consecutive periods. In the thesis, the dynamic DEA models are classified into dynamic DEA models with carry-overs (links) and dynamic DEA models without carry-overs. The thesis focuses on dynamic DEA without carry-overs, mainly on one-stage dynamic DEA models without carry-overs which compute the overall efficiency in one stage. The aim is to formulate original dynamic DEA models. I introduce the basic one-stage dynamic DEA model and prove that the basic one-stage dynamic DEA model gives the same results as the basic two-stage dynamic DEA model. I propose the advanced dynamic DEA models without carry-overs with the increasing vector, the convex vector, and the ratio scale vector of time weights. Further, I propose dynamic DEA models without carry-overs with normalised inputs and outputs with the increasing vector, the convex vector, and the ratio scale vector of time weights. The application part of the thesis analyses a dataset regarding 18 Dividend Aristocrats from the consumer sectors in the United States. The dataset consists of 16 consecutive years, from 2005 to 2020. Two inputs are used Market Capitalisation and Shareholders' equity in billions of USD, and two outputs, Earnings and the Total Dividends paid in millions of USD. The proposed models are computed for this dataset and are compared with the results of static models and already-known dynamic DEA models without carry-overs mentioned in this thesis. The models are also compared based on the time consumption for the computation. All calculations are carried out using our original procedures written in the Lingo modelling language.