As an extension of the neoclassical urban systems theory (Henderson, 1974), we develop a general theory of regional (inter-city) price dispersion which also explains the “subnational Penn effect,” i.e., cross-city correlations among population size, prices, real income and human capital stock. The model is also a theory of international price dispersion that is observationally equivalent to and more appealing than the Balassa-Samuelson theory, implying that the (international) Penn effect may simply be an aggregate result of the “subnational Penn effect.” Furthermore, it shows that, contrary to the popular view, economic integration can increase as well as decrease spatial price variation.