Does the monocentric model work in a polycentric urban system? An examination of German metropolitan regions

Abstract

In explaining the determinants of urban spatial structure, researchers have relied on the traditional monocentric ‘Alonso-Mills-Muth’ model. This article contributes to this discussion by testing the viability of the monocentric model when applied to metropolitan areas in Germany, a country traditionally associated with a polycentric urban structure, regional differences and urban shrinkage. We estimate the model with a unique dataset covering 92 metropolitan areas over two time periods (2000 and 2014), which allows estimation in both a cross-sectional and a panel framework. Using spatial and panel regression techniques, we test whether the underlying determinants of urbanisation vary according to factors unique to the German context, including the roles of historical geography, regional polycentricism and urban shrinkage. We found that, similar to the US studies, the model performed reasonably well, particularly with the overall fit and the performance of the population variable, which was significant and positively related to urbanised area. Personal income and land prices showed mixed results, and the coefficients for transportation costs proved to be challenging. We also found that regional geography matters: a region in eastern Germany is smaller than one in the west. A proxy variable for regional polycentricity was not significant. Finally, we found that the model’s behaviour differs between growing and shrinking regions, most notably in the differing impacts that population change has on the change in urbanised area.

Is the monocentric urban economic model still empirically relevant? Assessing urban econometric predictions in 192 cities on five continents

“…Third, as predicted by the SUM, higher agricultural land prices decrease urbanized areas. This result confirms those of Oueslati et al (2015) and Paulsen (2012), whereas Spivey (2008) and Schmidt et al (2020) found mixed or non-robust results.…”

“…Second, higher incomes also significantly increase urban areas, in line with the predictions of the SUM and of Paulsen (2012), Oueslati et al (2015), and Schmidt et al (2020), but in contradiction with Spivey (2008). Spivey (2008), working on US cities, find a negative and significant impact of incomes on urban areas, which she explains by the fact that the increase in incomes increases the demand for more affordable housing farther from the city center, but also raises the opportunity cost of time, the second effect outweighing the first one.…”

“…Our results are in line with existing studies, and confirm the empirical relevance of the predictions of the SUM on the size of cities. First, higher populations significantly increase urban areas, as it is expected (and as it was empirically assessed, for instance, in Paulsen, 2012;Oueslati et al, 2015;Schmidt et al, 2020;Spivey, 2008). We also find an elasticity of urbanized area with respect to population which is lower than one, as predicted by the SUM.…”

“…They find a positive coefficient as expected. Finally, Schmidt et al (2020) use commuting time as a proxy for the opportunity cost of transportation and average diesel prices per liter as a monetary cost measure, but find non-robust results or even not the expected sign. Our proxies are close to those of Schmidt et al (2020), but lead to significant and robust results for commuting speed, in line with the SUM.…”

“…Recent examples include Paulsen (2012), which shows that the main predictions of the SUM are valid in the United States, with population and income determining urban areas. Oueslati et al (2015) does a similar exercise for 282 European cities, Schmidt et al (2020) for 92 cities in Germany and Jedwab et al (2021) for 1010 cities in developing countries.…”