policymakers demonstrate

Skilled Cities and Efficient Urban Transport 385
centres). To a certain extent, the urban system reflects the administrative hierarchy
of territorial entities. Because public services are subject to different
degrees of increasing returns, cities accommodate a variable number of governmental
departments and agencies, hospitals, universities, museums, and the
like. More importantly, cities have a different industrial composition. In the
past, cities produced a wide range of goods that were not traded because shipping
them was expensive. Once transport costs decreased sufficiently, mediumsized
and small cities became specialized in the production of one tradable
good. This increased specialization often leads to significant labour productivity
gains but makes cities vulnerable to asymmetric shocks. Today, only a few
urban giants accommodate several, but not all, sectors.
Unlike specialized cities, diversified cities are better equipped when confronted
with asymmetric shocks. Besides spillover effects between sectors, the
coexistence of different sectors may also reduce the uncertainty associated with
the initial phases of the product cycle (Duranton and Puga, 2001). For example,
the preliminary stages in the development of a new technology or product
require repeated contacts among those involved, which are much easier when
these people are in close proximity. Information becomes a spatial externality
because, as it circulates within the local cluster of firms and workers, it inadvertently
contributes to aggregate productivity. However, as shown by Helsley
and Strange (2014), potentially beneficial clusters do not necessarily emerge,
while the co-agglomeration that does occur in diversified cities may not be that
which creates the greatest productive benefits.
Henderson (1974, 1988) has developed a compelling and original approach
that allows us to describe an urban system that involves an endogenous number
of specialized cities trading goods. The second-generation models explore both
the sorting of workers and the composition of population across cities, which
are consistent with recent empirical evidence (Behrens et al., 2014, Eeckhout
et al., 2014). Davis and Dingel (2015) observe that in the US the hierarchy
of skills is highly correlated to the urban hierarchy. Specifically, these authors
have proposed a new modelling strategy that suggests that ‘the most skilled
individuals in the population live only in the largest city and more skilled
individuals are more prevalent in larger cities’. What makes these new models
especially appealing is their ability to capture what we know from urban
economics about the role of human capital externalities in the formation of
cities.
However, in this strand of literature, cities produce the same good or, equivalently,
different goods traded at zero cost. These models do not recognize
that cities are anchored in specific locations and embedded in intricate networks
of trade relations that partially explain the cities’ size and industrial
mix. In sum, they put aside the fact that location matters because trade is
costly (see Chapter 8). Allowing for a large number of potentially asymmetric