The published empirical literature on frontier production functions is dominated by two broadly defined estimation approaches – parametric and non‐parametric. Using panel data on Korean rice production, parametric and non‐parametric production frontiers are estimated and compared with estimated productivity. The non‐parametric approach employs two alternative measures based on the Malmquist index and the Luenberger indicator, while the parametric approach is closely related to the time‐variant efficiency model. Productivity measures differ considerably between these approaches. It is discovered that measures of efficiency change are more sensitive to the choice of the model than are measures of technical change. Both approaches reveal that the main sources of growth in Korean rice farming have been technical change and productivity improvements in regions of the country that have been associated with low efficiency.