“…This is because the nonparametric DEA measures all departures from the frontier as inefficiency, whereas the parametric SFA allows for a random error term (see Cummins and Weiss, 2000). However, like other studies employing both DEA and SFA (see, e.g., Fecher et al, 1993), the implications that can be drawn from our SFA results are generally consistent with those that can be drawn from the DEA results. Average technical efficiency in life insurance (0.84) is again higher than in non-life insurance (0.81), although the difference is smaller compared to DEA.…”