AbstractIt is unusual for the precise price data required by economic theory to be available, and as a result, many applied studies use information on regional or average prices as proxies for the specific prices required. The use of one price as a proxy for another, however, introduces all the problems associated with the use of proxy variables into the analysis; and these problems may persist, particularly those associated with specification errors, even if the proxy is highly correlated with the true variable. It is possible, however, to minimize the impact of specification errors by adding additional variables.