Household perception bias on water price in China: Asymmetric impacts and policy treatment

• Published in: Water resources and economics Vol. 51; p. 100262
• Main Authors: Jia, Jun–Jun, Luo, Li, Jiang, Maorong, Wu, Huaqing
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.07.2025
• Subjects: Asymmetric impacts; Average water price; Binary machine learning classifiers; China; household surveys; Information treatment; issues and policy; Perception bias; prices; regression analysis; water; water conservation; Water-saving measures; China; Average water price; Asymmetric impacts; Information treatment; Perception bias; Water-saving measures; Binary machine learning classifiers
• ISSN: 2212-4284; 2212-4284
• DOI: 10.1016/j.wre.2025.100262

Residents tend to respond to perceived water price rather than the true water price when making household water consumption decisions. The paper estimates household perception bias on the average water price and explores its impact on the adoption of daily water-saving practices, by using the unique 5449 household survey data across 50 cities in China. The bias refers to the discrepancy between perceived price and the true average price. Results from the multi-level regression model show that households can hardly perceive the true average water price accurately. Approximately 71.5 % of households underestimate the true average price to varying degrees. On average, households underestimate the true average water price by 19.3 %, which is equivalent to 0.761 Yuan per ton. There are asymmetric impacts of household perception bias. For one thing, only the underestimation bias significantly impacts on the adoption of both technical and behavioral water-saving measures. For another, it hinders the adoption of technical measures among high water consumption households, while it impedes the adoption of behavioral measures among low water consumption households. Among a total of seven machine learning classification algorithms, the Random Forest binary classifier, based on ten easily-answered feature questions, demonstrates the best performance in identifying households with underestimation bias. It constitutes a promising policy tool to implement information treatment on households with underestimation bias. It can facilitate water conservation resulting from downward perception bias, particularly by tapping into the greater water-saving potential of technical water-saving measures and high water consumption households.