This article introduces a non-parametric (linear programming) approach to calculate an input-based Malmquist productivity index, which is applied to Swedish pharmacies from 1980 to 1989. The methodology combines ideas from Farrell's efficiency measurement and Caves, Christensen, and Diewert's productivity measurement. The authors use a piecewise linear model to distinguish between changes in efficiency and changes in the production frontier, allowing for inefficiencies. The study focuses on 42 group pharmacies, which are a small but representative sample of Swedish pharmacies. The results show that the proposed method yields different productivity changes compared to the current method used by Apoteksbolaget, the public monopoly responsible for pharmaceutical retail in Sweden. The current method has limitations, including assumptions about technology, inability to distinguish between efficiency and frontier changes, and the need for a priori chosen weights.This article introduces a non-parametric (linear programming) approach to calculate an input-based Malmquist productivity index, which is applied to Swedish pharmacies from 1980 to 1989. The methodology combines ideas from Farrell's efficiency measurement and Caves, Christensen, and Diewert's productivity measurement. The authors use a piecewise linear model to distinguish between changes in efficiency and changes in the production frontier, allowing for inefficiencies. The study focuses on 42 group pharmacies, which are a small but representative sample of Swedish pharmacies. The results show that the proposed method yields different productivity changes compared to the current method used by Apoteksbolaget, the public monopoly responsible for pharmaceutical retail in Sweden. The current method has limitations, including assumptions about technology, inability to distinguish between efficiency and frontier changes, and the need for a priori chosen weights.