Working Paper: NBER ID: w11339
Authors: Russell Hillberry; David Hummels
Abstract: A large literature has shown that geographic frictions reduce trade, but has not clarified precisely why. We provide insights into why such frictions matter by examining which parts of trade these frictions reduce most. Using data that tracks manufacturers' shipments within the United States on an exceptionally fine grid, we find that the pattern of shipments is extremely localized. Shipments within 5-digit zip codes, which have a median radius of just 4 miles, are 3 times larger than shipments outside the zip code. We decompose aggregate shipments into extensive and intensive margins, and show that distance and other frictions reduce aggregate trade values primarily by reducing the number of commodities shipped and the number of establishments shipping. We consider two broad reasons for these facts and conclude that trade in intermediate goods is the most likely explanation for highly localized shipments and the dominant role of the extensive margin. In addition, we find no evidence of state-level home bias when distances are measured precisely and trade is observed over a very fine grid.
Keywords: Trade; Geographic Frictions; Microdata
JEL Codes: F1; R3
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| trade in intermediate goods (F10) | localized nature of shipments (L87) |
| geographic frictions (R12) | number of shipments (L87) |
| distance (R12) | number of shipments (L87) |
| geographic frictions (R12) | trade (F19) |
| distance (R12) | likelihood of shipments occurring (L87) |
| regional characteristics (R11) | likelihood of shipments occurring (L87) |
| local demand (R22) | likelihood of shipments occurring (L87) |