Use the data in airfare1.dta for this question. Consider the following panel regression model

[1-1] log(fare_it) =q_t+b₀+b₁concen_it+b₂log(passen_it)+b₃log(dist_i)+b₄[log(dist_i)]²+v_i+e_it,

where fare_it is the average one-way fare, concen_it measures market concentration (proxied by the market share of the biggest carrier), dist_i is the distance in miles and passen_it denotes the average number of passengers per day. Note that q_t means allowing for different year intercepts ( to 2000) and the unobserved effect v_i contains characteristics of each of the 1149 flight routes.

• Report and interpret the RE estimation result on equation [1-1], being sure to include year dummies.

Name a couple of characteristics of a route (other than distance). Discuss if these characteristics might be correlated with concen_it and/or passen_it.

Report and interpret the FE estimation result on equation [1-1]. Compare the result with that obtained in (a). Perform the Hausman Test (FE vs. RE).

Report and interpret the Hausman-Taylor estimation result on equation [1-1], treating “log(passen_it)” as the only endogenous variable in the regression.

Report and interpret the CRE estimation result on [1-1].

Attachments: