When Agglomeration Theory Meets Development Reality: Preliminary Lessons from Twenty World Bank Private Sector Projects

The World Bank has currently committed $1.5 billion to various projects that promote agglomeration benefits across firms, mostly in sub-Saharan Africa. Agglomeration benefits arise from the sharing of knowledge and ideas, the availability of a larger pool of labor, and from lower costs of production and transportation due to the clustering of firms. This paper reviews information from 20 World Bank agglomeration-focused projects such as special economic zones, growth poles, and industrial clusters. While many are at an early stage, it is possible to draw some tentative conclusions. First, we find that project designs overestimate the intellectual clarity of the agglomeration approach. There have been many implementation challenges, resulting in unsatisfactory ratings for several projects. We argue that greater efforts should be made to reduce technical complexity (including the number of activities and institutions involved), to build institutional capacity, and to make projects flexible. Looking forward, agglomeration projects may benefit from being designed as a series of incremental projects, by which lending can be carried out in stages to reflect changing realities on the ground. Second, there are important second-order effects, including the impact of projects on asset prices, notably land which needs to be incorporated into assessing project outcomes. Agglomeration projects and local investment can have a large effect on land values, raising issues of ownership, control, and allocation. Finally, while agglomeration may ease coordination failures and induce a critical mass of private firms to enter and invest, this does not always occur. Some threshold (in terms of lower costs and risk) may need to be reached to spur private investment. This threshold concept has not been incorporated into the current linear results framework (input to output to outcome); further experimentation is needed to develop flexible results frameworks that explicitly recognize this threshold effect.