Evaluating healthcare system efficiency of OECD countries: A DEA-based study

We evaluate healthcare system efficiency of 34 Organisation for Economic Cooperation and Development (OECD) member countries using Data Envelopment Analysis (DEA). Our implementations rely on data for the years 2008 and 2012. Our base model is an output oriented Banker-Charnes-Cooper model that uses the number of physicians, nurses, beds per 1000 population as inputs, and life expectancy at birth, infant survival rate as outputs. We observe that the distinction between efficient/inefficient countries is not necessarily in line with the classification of developed/developing countries, which is consistent with findings in past studies. We then build Assurance Region Global (ARG) versions of our base model in order to impose restrictions on weights assigned to outputs, thus limiting DEA’s inherent ability to interpret a decision making unit’s performance from a most optimistic point of view for that particular unit. We observe that some developing countries remain efficient in these models as well. Analysis of the results of our ARG model with 2008 data shows that Luxembourg appears as a reference for inefficient countries most frequently whereas Sweden, Chile and Spain appear in the reference list multiple times. In the 2012 ARG model, Slovenia appears most often as a reference followed up by Iceland and Spain. We then analyze efficiency with respect to the alternative output of survival from most likely causes of death such as ischemic heart disease, cerebrovascular diseases and malignant neoplasms (cancer) on 2008 and 2012 data. France is the model country for most of the inefficient countries with the new output variables. While there are some differences between the efficient countries of the base model and the modified one, the overlap is quite strong.