UC Berkeley

As developed economies have shifted from producing manufacturing goods, to been producers of knowledge and innovation, the traditional drivers of economic growth are giving way to an economy based on skills and creativity. As a result, the aptitudes and knowledge of the workforce have become important drivers for economic growth. The importance of geography and the forces of agglomeration in determining the location of human capital will keep growing, and cities with a large percentage of highly skilled workers will become the focal points for this transformation.

This new industrial revolution has exacerbated regional disparities by an order of magnitude. Cities are diverging not only in their housing prices and productivity, but also in the skill composition of their workforce. Inequality has not only increase between regions but important changes have also taken place within regions, as large cities have become more unequal than the countries that host them.

In this context it is more important than ever to understand how cities operate, and what drives the value of proximity in a knowledge economy. This is complicated since the uneven distribution of economic activity in space is partly driven by endogenous interactions between firms and workers in goods and factor markets which can move between regions.

In Chapter I, titled "On the Geography of Inequality: Labor Sorting and Place-Based Policies in General Equilibrium", I study how city fundamentals, like amenities and housing restrictions, contribute to aggregate wage inequality through the sorting of heterogeneously skilled workers. I develop a ``system of cities'' model that features workers who differ along a continuum of skills and who compete for limited housing. This model is quantitatively tractable, and can replicate patterns in the dispersion of wages and housing prices both between and within cities. I calibrate this model to match different moments of the distributions of talent and wages for a cross-section of US cities, and I use it to understand the importance of sorting when accounting for patterns of regional inequality. I then evaluate the general equilibrium effects of an important place-based policy, namely housing policy, and find that a 1\% expansion in the supply of houses in more constrained cities can improve aggregate productivity between 0.2\% and 0.4\%. These effects would be larger in the absence of sorting. Moreover, relaxing housing constraints in those cities also tends to increase aggregate wage inequality.

In Chapter II, titled "Urban Connectivity", I study how technological changes that affect urban connectivity (the efficiency with which workers use their productive time in a city) can explain the increased spatial segregation in workers' skills and firms' productivity. I focus in an economy that produces knowledge and requires the matching of heterogeneous firms and workers. I provide a spatial equilibrium model that has the unique feature that allows for the sorting of a continuum of firms and workers where productive complementarities are city specific. I show that small changes in the connectivity of a city, can generate non-linear changes in city sizes and the level of skill segregation between cities. This suggest that small shocks to the productive environment of a city could account for the important changes we have observed in workers' skills and firms' productivity distributions.

Finally, in Chapter III, title “Clustering to Coordinate: Who Benefits From Knowledge Spillovers?”(joint work with William Grieser and Gonzalo Maturana), we study location and investment decisions by firms. Firm clustering is positively correlated with productivity, and it exhibits significant cross-sectional variation across industries. Thus, it is important to understand the industry characteristics that drive firms' decisions to co-locate. We develop a model of knowledge sharing and derive the prediction that riskier and more complex industries experience the greatest gains from knowledge spillovers. Using tests that account for the non-randomness of location decisions, we find a strong positive relationship between industry risk or complexity and the clustering of: 1) firms' headquarters, 2) patent inventors, and 3) R