Many nonmarket valuation models, such as the Ricardian model, have been estimated using cross sectional methods with a single year of data. Although multiple years of data should increase the robustness of such methods, repeated cross sections suggest the results are not stable. We argue that repeated cross sections do not properly specify the model. Panel methods that correctly specify the Ricardian model are stable over time. The results suggest that many cross sectional methods including

more »

... nic studies and travel cost studies could be enhanced using panel data. We are grateful to William Nordhaus for comments on an earlier draft of this paper. Emanuele Massetti gratefully acknowledges financial support from the EU sponsored PASHMINA project and by the research project D.1 -2010, Metodi e modelli matematici per le decisioni nelle scienze sociali, Catholic University of Milan. Many nonmarket valuation models, such as the Ricardian model, have been estimated using cross sectional methods with a single year of data. Although multiple years of data should increase the robustness of such methods, repeated cross sections suggest the results are not stable. We argue that repeated cross sections do not properly specify the model. Panel methods that correctly specify the Ricardian model are stable over time. The results suggest that many cross sectional methods including hedonic studies and travel cost studies could be enhanced using panel data.