Paid for Size, Not Hardship: Diagnosing and correcting implicit structural compensation in Swedish electricity distribution benchmarking

Abstract

Sweden’s energy regulator uses DEA to benchmark 144 electricity distribution operators against five outputs. One of these, network stations (NS), is intended to compensate firms for spatially dispersed networks. We show this compensation is systematically misallocated. At the aggregate level, NS compensation tracks density; at the firm level, a subset of predominantly urban operators, including the three largest DSOs, receive NS credit that their density does not predict. Routing density through StoNEZD’s parametric z-channel corrects efficiency ratings for over-compensated firms and produces a statistically significant shift. Moreover, the current specification’s exposure to the curse of dimensionality is far narrower than its five-output count implies. The effective rank of the output cloud is approximately 1.12, a single scale-axis with NS as the lone orthogonal direction. The curse of dimensionality therefore reduces to the curse of NS.