A brief review of the literature on how changes in transport costs can change the spatial makeup of economies. Consider a world with positive transport costs between regions, and constant returns to scale technology: there is no reason for trade, since transport is costly, and everything can be produced at the same cost in each region. Therefore, each household will be an autarky. To generate cities and trade, then, we need some sort of increasing returns to scale in production, whether than be directly in the factor, through some aggregation of knowledge, or whatever. Given that premise, nonobvious results are widespread in this theory. For instance, imagine a poor and a rich country decrease the transport costs between the two (i.e., the EU builds a railroad from Germany to Poland). In a spatial model, this will generally lead to more aggregation of high-value production in the wealthier country, due to greater efficiency in the high-value sector resulting from increasing returns to scale.
Clearly, production is not the only reason cities exist – Glaeser, among others, has pointed out that consumption externalities, like access to opera, exist as well – but nonetheless understanding the economics of cities in a world of declining transport and communication prices is critical for current urban policy.