Variable returns to agglomeration and the effect of road traffic congestion

This paper investigates the links between returns to urban density, productivity and road traffic congestion. A generalised translog production-inverse input demand function is estimated to test for the existence of variable returns to agglomeration in manufacturing, construction and service industries. Two separate measures of urban density, in which proximity is represented by straight line distance or by generalised cost, are constructed and included in the translog to identify the effect of road traffic congestion. The results show that for some sectors of the economy diminishing returns to urban density can set in causing the magnitude of agglomeration elasticity to fall as effective densities increase. The generalised cost based measure of agglomeration produces higher elasticities because it captures both time and distance dimensions of density. A comparison of spatial variance in estimates indicates that road traffic congestion plays an important role in explaining diminishing returns for the most highly urbanised locations.