WHAT CAUSES SPATIAL DISPERSION OF ECONOMIC ACTIVITY?*
By Pedro Herrera Catalán
[Click Here To Download The Paper]
View map
The study of the impact of transport costs and barriers to trade on the agglomeration process of manufacturing sector have habitually been the main concern of the Economic Geography literature. Unfortunately, the agricultural sector has comparatively received little attention regarding its impact on industry localisation. According to Takatsuka and Zeng (2012) this is mainly due to the simplifying assumptions of formulation models, which usually consider the agricultural goods to be homogenous and transportable at zero cost in order to offset the trade imbalance in the manufacturing sector.
This assumption however is empirically unsatisfactory in the real world since agricultural transport costs play a key role when defining the spatial configuration of economic activity, especially in developing countries where an important part of their inhabitants is devoted to agricultural activities (Picard and Zeng, 2005).
A few Economic Geographers have researched the role of the transport costs of the agricultural good in core-periphery models. Nevertheless, no consensus had emerged about the impact of agricultural transport costs on agglomeration process. Fujita et al. (1999) find that an increase in agricultural transport costs causes dispersion as intensely as a rise in manufacturing transport costs. Davis (1998) finds that the assumption of free transport of the agricultural good is not inoffensive because dispersion emerges if the agricultural good is transported with identical positive cost as the manufactured goods. Yu (2005) finds that industry agglomeration may either disappear, be reversed, or remain when the transport costs of the agricultural good are positive.
This paper examines the impact of agricultural transport costs upon the firms’ incentives to disperse or to agglomerate in agriculture-related industries in Peru. By doing so, we employ a combination of nonparametric and parametric methods. First, using a nonparametric procedure developed by Duranton and Overman (2005), we find that agriculture-related industries do not exhibit strong localisation patterns. Only 1 out of 13 industries in our sample is highly localised at short distances whereas the remaining industries are dispersed or are localised at long distances with a significantly weaker degree of intensity.
We then explore cause-effect linkages between agricultural transport costs and industrial location using parametric conditional logit models. We find that the probability that a rural micro-geographic area has a firm highly localisation at short (long) distances is higher if it has low (high) agricultural transport costs. Likewise, weak localisation at long distances and dispersion patterns are explained by the interplay of high and low transportation costs in the agricultural sector. Thus, both high and low levels of agricultural transport costs operate as dispersion forces that ultimately cause dispersion of economic activities in agriculture-related industries.
These results argue in favour of the branch of the Economic Geography literature that supports the positive relationship between agricultural transport costs and dispersion of economic activities.
* Paper presented at 61st NARSC Congress, Washington, D.C., EEUU, November 12-15, 2014