Nice paper by Thomas Klier and Dan McMillen of Chicago Fed.
They probe the reasons for clustering of auto production in Europe:
Motor vehicle and motor vehicle parts production plants tend to exhibit a strong degree of agglomeration. This paper estimates a spatial model utilizing detailed plant-level data that is pooled across seven countries in Europe. The paper makes several contributions.
First, we assemble a set of nearly 1,800 European plant locations of the largest motor vehicle parts suppliers, as well as the location of all light vehicle assembly plants operational in 2010.
Second, we obtain detailed spatial data – at a higher resolution than what is provided by the NUTS-3 regions – for five European countries (France, Spain, Italy, Poland, and the Czech Republic). For the U.K. (ward level) and Germany (community level) we acquired spatial data at an even more detailed level. These seven countries are home to over 70% of the plants in our data set. The ability to pool data from multiple countries allows us to estimate a location model for a large share of the vehicle parts industry, an industry that extends across all of Europe.
The modeling results suggest that the main forces of agglomeration in the European auto supplier sector are (1) highway access (connecting supplier plants as well as suppliers and their downstream customers, the assembly plants), (2) the desire to locate near assembly plants, (3) as well as near other parts producing plants.
As econ geographers would predict…