4.2. Industrial Spatial Structures
The first estimated model (Table 6) identifies the explanatory variables that are relevant to
major industrial agglomerations. These agglomerations are measured based on the IVA of each
municipality. Significant variables in explaining such agglomerations were: QLA, QLC, POP, BI,
BCD, BCND and CTSPM. In addition, the spatial lag model was found to be the most adequate in the
specification tests.
The positive and significant value of the coefficient of the lagged dependent variable (W_IVA)
is no reason to rule out the hypothesis of a global spatial autocorrelation in the explanatory variables
and error terms10. This implies that changes (shocks) associated to both the variables that have been
added and those that have not been added to the model will produce effects causing a municipality’s
features to spill over into its neighbors. These effects are most noticeable to the closest neighbors,
becoming increasingly less noticeable as one moves away from their source.
It is no surprise that the resident population of the municipality (POP) and its surrounding area
proves to be the most statistically significant variable in explaining the local industrial agglomeration
level. This is a proxy variable of the urban scale that is usually found in literature. It confirms the
significance of diversification or Jacobian external economies, stemming from the urban scale, in
attracting and agglomerating industrial activities (Pred, 1966; Jacobs, 1969; Glaeser et al, 1992). Upper
schooling (E25) and infrastructure (ESGT) variables were not significant, and the same is true for the
dummy variable for non-metropolitan municipalities.
Sectoral variables BI, BCD and BDND capture the influence of the municipality’s sectoral
structure on industrial concentration as measured by the IVA. Results indicate that the municipalities
with larger presence of companies producing capital and durable goods have a larger IVA, while
municipalities with prevalence of manufacturers of non-durable consumer goods have a smaller IVA.
This relationship was somehow expected: major value-added industrial agglomerations are comprised
of world competitive companies that are capable of differentiating themselves technologically and
involve directly or indirectly manufacturers of capital and durable goods (A type companies); these are
“polarizing” companies. The opposite is usually true for non-durable consumer goods industries:
companies are not very competitive and use established technologies. Such companies will not give
rise to industrial agglomerations and, as a matter of fact, they tend to be located outside of these
agglomerations.
10 The spatial weight matrix used in this study is a contiguity matrix for the 5,507 municipalities, built using ArcView 3.2,
pursuant to Queen criteria. A matrix was built for the distances between the seats of the various municipalities, but it was
not possible to use it with the models because of the computer’s storage capacity and the file’s size (1.2GB).
17
More intriguing information
1. The name is absent2. On the job rotation problem
3. National curriculum assessment: how to make it better
4. The name is absent
5. Insecure Property Rights and Growth: The Roles of Appropriation Costs, Wealth Effects, and Heterogeneity
6. The name is absent
7. News Not Noise: Socially Aware Information Filtering
8. AGRICULTURAL PRODUCERS' WILLINGNESS TO PAY FOR REAL-TIME MESOSCALE WEATHER INFORMATION
9. The mental map of Dutch entrepreneurs. Changes in the subjective rating of locations in the Netherlands, 1983-1993-2003
10. Learning and Endogenous Business Cycles in a Standard Growth Model