DEA와 SFA를 활용한우체국보험의 효율성 평가에 관한 연구

The paper evaluates the efficiency of the Korean postal insurance and also identifies important factors affecting the outputs in order to provide the post office with valuable direction of the business innovation. In the paper, both premium income and agent income of the regional post offices are employed as output variables while market orientation, self-effort and the number of agents, their ages and duration of jobs are used as input variables. To measure efficiency, we apply not only data envelopment analysis (DEA) as a non-parametric method but also stochastic frontier analysis (SFA) as a parametric method. In particular, the paper focuses on analyzing error terms which can be decomposed into random and efficiency errors in order to check whether there exists significant difference in efficiency among postal offices. We also employ Cobb Douglas multiplicative models for log-transformed data to reflect the relationship between the original input and output variables considering the law of diminishing return. The empirical results using data of total 120 postal offices reveal that for premium income as output, the model errors are mainly due to random errors rather than efficiency errors. That is, premium income model is fully explained by the input variables. On the other hand, for agent income as output, the model errors come from both random and efficiency errors but the latter is much bigger, which indicates there exists difference in efficiency. Namely, agent income is not fully explained by the input variables. Another important finding is that self-learning efforts of agents significantly affect their income. In addition, market orientation of the post offices also works for improvement of the outputs for the group maintaining a higher level of market orientation. So we recommend the postal insurance to enhance market orientation to a certain threshold and also encourage the agents to make self-learning efforts.