How Should We Measure City Size? Theory and Evidence Within and Across Rich and Poor Countries

It is obvious that holding city population constant, differences in cities across the world are enormous. Urban giants in poor countries are not large using measures such as land area, interior space or value of output. These differences are easily reconciled mathematically as population is the product of land area, structure space per unit land (i.e., heights), and population per unit interior space (i.e., crowding). The first two are far larger in the cities of developed countries while the latter is larger for the cities of developing countries. In order to study sources of diversity among cities with similar population, we construct a version of the standard urban model (SUM) that yields the prediction that the elasticity of city size with respect to income could be similar within both developing countries and developed countries. However, differences in income and urban technology can explain the physical differences between the cities of developed countries and developing countries. Second, using a variety of newly merged data sets, the predictions of the SUM for similarities and differences of cities in developed and developing countries are tested. The findings suggest that population is a sufficient statistic to characterize city differences among cities within the same country, not across countries.