Working Paper: NBER ID: w28772
Authors: Benjamin Schoefer; Oren Ziv
Abstract: Why do cities differ so much in productivity? A long literature has sought out systematic sources, such as inherent productivity advantages, market access, agglomeration forces, or sorting. We document that up to three quarters of the measured regional productivity dispersion is spurious, reflecting the “luck of the draw” of finite counts of idiosyncratically heterogeneous plants that happen to operate in a given location. The patterns are even more pronounced for new plants, hold for alternative productivity measures, and broadly extend to European countries. This large role for individual plants suggests a smaller role for places in driving regional differences.
Keywords: Productivity; Urban Economics; Granularity Bias
JEL Codes: D22; D24; E23; R0; R12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| granularity bias (D80) | spurious productivity differences (O49) |
| idiosyncratic plant productivity differences (D29) | inflated perceptions of productivity differences (D29) |
| granularity bias (D80) | variance of average productivity levels across U.S. cities (R12) |
| raw variance of average productivity levels (O47) | true productivity differences attributable to place (O49) |
| luck of the draw of plant distributions (D39) | productivity differences (O49) |