Working Paper: NBER ID: w4730
Authors: Vernon Henderson
Abstract: Using a panel data set of county-level employment in machinery, electrical machinery, primary metals, transportation, and instruments, this paper analyzes the role of dynamic externalities for individual industries. Key issues examined include the role of externalities from own industry concentration (localization, or MAR externalities) versus the role of externalities from overall diversity of the local environment (urbanization, or Jacobs externalities). In contrast to previous studies, use of panel data allows us to separate these effects out from fixed/random effects influencing industries over time. Panel data also allow us to estimate a lag structure to externality variables, indicating how long history matters and the time pattern of effects. A particular issue concerns whether conditions from the immediate year or so prior to the current have the biggest impact on current employment, or periods several years prior have the largest impact. For all industries both localization and urbanization effects are important. For traditional industries most effects die out after four or five years, but for high tech industries effects can persist longer. The biggest effects are typically from conditions of three to four years ago, in the county and metropolitan area.
Keywords: externalities; industrial development; dynamic externalities; employment; localization; urbanization
JEL Codes: R11; L52
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| historical industrial environment (3-4 years prior) (N63) | current employment levels (J63) |
| past employment levels (J63) | current employment levels (J63) |
| localization effects (MAR externalities) (F69) | current employment levels (J63) |
| urbanization effects (Jacobs externalities) (R11) | current employment levels (J63) |
| historical conditions of local industrial environment (N93) | current employment outcomes (J68) |
| effects for traditional industries dissipate after four to five years (L16) | current employment levels in traditional industries (J21) |
| high-tech industries exhibit longer-lasting impacts from historical conditions (L63) | current employment levels in high-tech industries (L63) |