DP17126 A granular spatial model
We develop a granular spatial model (GSM) which introduces indivisible workers and firms into the canonical quantitative urban model. Assortative matching between heterogeneous employers and employees leads to the formation of large granular firms, which generates realistic agglomeration patterns with many empty locations and a few extremely dense ones, even with flat location fundamentals. This feature makes the GSM particularly suitable for the study of multiple equilibria and the long-run effects of temporary spatial shocks such as natural disasters or place-based policies. As almost all firms are inframarginal in the GSM, especially the most productive ones in the most dense locations, place-based policies become `place-based lotteries': their expected payoff critically hinges on the probability to attract a few large firms. We illustrate these insights using as an example the spectacular recent rise of Chicago's Fulton Market district west of the Loop.