Abstract:
ABSTRACT The present paper shows that product‐specific spatial price variation usually causes the Country‐Product‐Dummy (CPD) method to be biased. In cases where it is not, the CPD method is still inefficient and statistical inference is invalid. The paper develops a non‐linear generalization of the CPD method. Even for product‐specific spatial price variation, this NLCPD method remains unbiased and allows for inference. A comprehensive simulation reveals that the NLCPD method's root mean squared error is smaller than that of the Gini‐Eltetö‐Köves‐Szulc (GEKS) index and the CPD method. Finally, this paper applies the NLCPD method to spatial price information derived from Germany's consumer price index micro data of 2019. Price levels of the 401 German districts are computed.