Working Paper: CEPR ID: DP6053
Authors: Kathryn Graddy; Peter E. Kennedy
Abstract: When a supply and demand model is recursive, with errors uncorrelated across the two equations, ordinary least squares (OLS) is the recommended estimation procedure. Supply to a daily fish market is determined by the previous night?s catch, so this would appear to be a good example of a recursive market. Despite this, data from the Fulton fish market are treated in the literature, without explanation, as coming from a simultaneous-equations market. We provide the missing explanation: inventory changes, influenced by current price, affect daily supply. Instrumental variable estimates using the full data set differ very little from OLS estimates using only observations with little inventory change, providing strong support for our explanation. Finally, we note that because of inventory changes, estimates of supply price elasticities in high-frequency markets must be interpreted with care.
Keywords: demand estimation; fish; Fulton market; inventories; simultaneous equations
JEL Codes: C3; L6
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| previous night's catch (wt-1) (Y10) | supply on day t (qt) (C69) |
| change in inventory (pt - pn) (D25) | supply on day t (qt) (C69) |
| current prices (P22) | inventory changes (G31) |
| inventory changes (G31) | supply on day t (qt) (C69) |
| current prices (P22) | supply on day t (qt) (C69) |
| supply on day t (qt) (C69) | inventory changes (G31) |
| supply on day t (qt) (C69) | previous night's catch (wt-1) (Y10) |