A review and methodological comparison of parametric and non-parametric approaches to measuring frontier efficiency for life insurers: evidence from a simulated dataset

摘要

This paper compares the multiple frontier efficiency techniques from parametric and non-parametric approaches using a simulated dataset of German life-insurers and develops two important research questions by conducting an extensive review of the literature. Using a two-input two-output production framework, five different frontier efficiency estimation techniques are considered which are stochastic frontier analysis (SFA), distribution-free approach (DFA), data envelopment analysis (DEA)-CCR, DEA-BCC, and DEA-RAM. The first contribution of this study is a research framework that compares multiple frontier efficiency techniques for life-insurers, which is rare given the exponential growth of efficiency studies in this sector. The second contribution of this study is use of a simulated panel dataset that generated well-behaved production data for estimation of frontier techniques. The findings suggest that, as opposed to previous studies, parametric and non-parametric approaches can both generate comparable technical efficiency scores with simulated data. Moreover, techniques from parametric approaches, that is, SFA and DFA, are consistent with each other, whereas the same applies to non-parametric approaches, that is, DEA models.