in urban areas. More recently, Ciccone and Peri (2000) find that a one-year increase in
average schooling in US cities raises aggregate productivity by 8 to 11%.6
How then, are externalities transmitted across regions? Is there any evidence that their
effects are weaker over longer distances? Ciccone and Hall (1996) use data at the county
level to see if variations in population density can explain the large discrepancies in
productivity levels across the US.7 In their analysis, they control for the endogeneity of
employment density to ensure that the correlation between density and productivity that they
find is not merely the result of productive regions growing faster than less productive ones.
They find that doubling employment density increases labour productivity by 6%. Their
results hold even when other factors are taken into account (such as the level of public
capital, the level of education and the influence of market size). According to their analysis,
closer interaction between workers in a geographical unit does have a positive effect on
productivity.
Other indirect evidence supports the argument that the relevance of spatial influence
decay as distance increases. In fact, the geographical distribution of income and
unemployment, both in absolute values and variations over time is strongly correlated across
neighbouring regions (see Overman and Puga, 1999, and Quah, 1996).
3. THE MODEL
We consider a world economy consisting of two regions (1,2). There are two sectors:
agriculture and manufacturing. Agriculture is perfectly competitive and produces a
homogeneous good according to a constant-return-to-scale technology. This sector employs
6 For a survey on empirical work relating schooling to aggregate labour productivity see de la Fuente and
Domenech (2000). They also estimate the effect of a one-year increase in average schooling on the average
labour productivity for OECD countries, finding evidence of a positive impact by around 4%.