Is there a ‘price’ to pay for agricultural TFP measurement? Limitations of the distance function approach

Abstract

Cross-country comparisons of agricultural total factor productivity (TFP) diverge markedly depending on method: superlative indices (e.g., Törnqvist) leverage price and quantity data, while quantity-only indices (e.g., Malmquist) rely solely on quantities, yielding inconsistent estimates. We theoretically demonstrate that this disparity stems from measurement errors in the quantity-only approach's implicit shadow prices, which deviate substantially from market prices employed by superlative methods, introducing noise and bias. Utilising a novel, cross-country consistent dataset of agricultural production accounts for the United States, Canada and Australia (1961–2006), we empirically affirm that the superlative index consistently outperforms its quantity-only counterpart in accuracy and aggregation stability across scales. This superiority, rooted in price data's capacity to reflect economic realities (e.g., input cost shifts), underscores the critical need for comprehensive price information in international productivity assessments. Our findings offer actionable guidance for agricultural economists and policymakers prioritising robust TFP metrics.