Reconsider the example used in Section 11,7 on the supply and demand for fish at the Fulton Fish Market. The data are in the file fultonfish.dat. In this exercise we explore the behavior of the market on days in which changes in fish inventories are large relative to those days on which inventory changes are small. Graddy and Kennedy

(2006) anticipate that prices and quantities will demonstrate simultaneity on days with large changes in inventories, as these are days when sellers are demonstrating their responsiveness to prices. On days when inventory changes are small, the anticipation is that feedback between prices and quantities is broken, and simultaneity is no longer an issue.

(a) Use the subset of data for days in which inventory change is large, as indicated by the variable CHANGE ¼ 1. Estimate the reduced-form (11,16) and test the significance of STORMY. Discuss the importance of this test for the purpose of estimating the demand equation by two-stage least squares.

(b) Obtain the least squares residuals ^vt2 from the reduced-form equation estimated in (a). Carry out a Hausman test8 for the endogeneity of ln(PRICE) by adding ^vt2 as an extra variable to the demand equation in (11,13), estimating the resulting model by least squares, and testing the significance of ^vt2 using a standard t-test.

If ^vt2 is a significant variable in this augmented regression then we may conclude that ln(PRICE) is endogenous. Based on this test, what do you conclude?

(c) Estimate the demand equation using two-stage least squares and ordinary least squares using the data when CHANGE ¼ 1, and discuss these estimates.

Compare them to the estimates in Table 11,5.