Working Paper: CEPR ID: DP17749
Authors: Kristian Behrens; Florian Mayneris; Théophile Ndjanmou Bida
Abstract: We construct a dataset on the size of parcels and on the footprintof buildings occupied by manufacturing establishments in Canada. Wefind that establishments in big cities and close to city centers occupytheir parcels more densely, both in terms of employment and buildingfootprint. Moreover, establishments with more employees use less landper worker. These facts cannot be rationalized by a Cobb-Douglas productionfunction featuring land and labor as production factors. Theyare, however, consistent with a modified-CES production function inwhich land has a fixed cost component. The calibrated elasticityof substitution between land—as measured by parcel size—and labor islow (between 0.2 and 0.4).
Keywords: land use; land and production; production function; georeferenced data; building and parcel polygons; manufacturing
JEL Codes: R32; R14; L60
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| low elasticity of substitution between land and labor (J49) | land and labor not easily interchangeable (Q15) |
| larger urban areas (R11) | higher employment density (R23) |
| distance from city centers (R12) | lower structural density (L74) |
| size of establishment (L25) | less land per worker (Q15) |
| city population (R23) | parcel size per worker (J29) |
| distance to city centers (R12) | parcel size per worker (J29) |