Reading "Agglomeration Economics", by Edward Glaeser

Agglomeration economics are the benefits that come when firms and people locate near one another together in cities and industrial clusters. These benefits all ultimately come from transport costs savings: the only real difference between a nearby firm and one across the continent is that it is easier to connect with a neighbor. Transport cost must be interpreted broadly, and they include the difficulties in exchanging goods, people and ideas. The connection between agglomeration economies and transportation costs would seem to suggest that agglomerations should become less important, as transportation and communication cost have fallen. Yet, a central paradox of our time is that in cities, industrial agglomerations remain remarkably vital, despite ever easier movement of goods and knowledge across space.

If transportation costs are so low, then why has the urge to agglomerate remained so strong?

Measuring agglomeration: price, wage, quantities

Urban economists infer urban success from high local wages, robust real estate prices, and growth in the number of people within an area. If a place is doing well, then employers should be willing to pay more for workers in that area, people should be willing to pay more for access to that place, and more people should move to that area. The logic of the spatial equilibrium is that since people can move freely within a nation, they must be indifferent between different locales. This indifference implies that high wages must be offset by high prices or low amenities; otherwise, people would flock to high-wage areas. High housing prices reflect high wages, high amenities, or both.

However, the spatial equilibrium concept only gives us one-half of the labor market equilibrium that determines area wages. The other half is labor demand –the willingness of firms to pay for their workers. So, shile high wages must reflect something bad about an area, like high prices or poor amenties, high wages must also reflect something good about an area that makes firms willing to tolerate a high cost of labor. Firms wouldn’t continue to locate to NY or the San Francisco Bay region unless those areas were productive enough to offset the cost of expensive workers.

Neoclassical economics tells us that wages reflect the marginal product of labor. In standard Cobb-Douglas formulation of the producer’s problem, where most capital is mobile, the high marginal product of labor in a given area must either reflect a high productivity level or an abundance of non-traded capital inputs to production. Wages, therefore, can be interpreted as telling us about the core determinants of urban productivity, and high wages in an area are usually interpreted as meaning that the area is unusually productive.

One of the facts that supports the existence of agglomeration economies is the strong relationship between density and high wages. This fact is mirrored in the strong relationship between area density and per capita gross metropolitan product (GMP). The fact is quite statistically robust, but the causal chain in the relationship is difficult to infer. Does the density-productivity relationship mean that the dense place become more productivity or that productive places attract more people? The need to tease out the direction of causality in this relationship motivates the first chapter in this volume.

One possibility is that dense places are more productive because they attract more skilled workers. Glaeser and Mare find little evidence that this is the case in US cities, but the selection of the skilled into cities seems to be much stronger in France.

Chapter 1. Estimating agglomeration economies with History Geology and worker effects. --- Pierre-Philippe Combes, Gilles Duranton, Laurent Gobillon, and Debastien Roux

Productivity and wages are higher in larger cities and denser areas. This fact was first noted by Adam Smith (1776) and Alfred Marshall (1890) and has been confirmed by the modern empirical literature on this topic (see Rosenthal and Strange, 2004 for review). The measured elasticity of local productivity with respect to employment density is typically between 0.04 to 0.10. (fig 1 show the positive relationship between mean log wages (also TFP) and employment density.)

Two fundamental identification problems must be dealt with. First, density and measure of productivity (wage or TFP) may be simultaneously determined. This could happen because more productive places tend to attract more workers and as a result become denser. And alternative explanation, albeit equivalent form tan econometric perspective, is that there may be a missing local variable that is correlated with both density and productivity. We refer to this issue as the endogenous quantity of labor problem. Since Ciccone and Hall (1996), a standard way to tackle this problem has been to use instrumental variables (IV).

The second major identification problem is that more productive workers may sort into denser areas. This may occur for a variety of reasons. For instance, skilled workers may have a stronger preference for high density, perhaps because density leads to better cultural amenities. Alternatively, the productivity benefits of high density may be stronger for skilled workers. These explanations suggest that it is not only dneisty that we expect to be simultaneously determined with productivity but also the characteristics of the local workforce. To make matter worse, we expect characteristics that are not usually observed by the statistician, such as ambition or work discipline, to matter and to be spatially unevenly distributed. For instance, French university professors may have similar observable characteristics everywhere, but a disproportionate fraction of the better ones are working in or around Paris. We refer to this problem as the endogenous quality of labor problem. Since Glaeser and Mare (2001), a standard way to tackle this problem has been to use the longitudinal dimension of the data.

One may also be concerned that density affects productivity in a myriad of ways, directly and indirectly (see Duranton and Puga, 2004). Denser markets allos for a more efficient sharing of indivisible facilities (local infrastructure), risk, and the gain form variety and specialization. Next, denser markets also allow for a better matching between employers and employees, buyers and suppliers, partners in joint projects, or entrepreneurs and financiers. This can occur through both a higher probability of finding a match and a better quality of matches when they occur. Finally, denser markets can facilitate learning about new technologies, market evolutions, or new forms of organization. Some of these mechanisms (matching) may have instantaneous effects, while other (learning) may take time to materialize.