Question

. One of my recent papers examine important and timely research questions using a field experiment approach in eBay auctions: (i) Can merchandise return policy (MRP; liberalness in the MRP) increase consumers’ willingness to pay? and (ii) is the marginal impact of MRP diminishing? In this study we created three brand new eBay seller profiles, all with zip-codes located within five miles of each other in a college town in the U.S. The eBay stores received exactly the same product description, pictures, outbound shipping policies, etc. The only difference among the three sellers was the extent of liberalness in the MRP and we chose to operationalize MRP liberalness in terms of the time window during which the customer is allowed to return the purchased product. The most conservative MRP (Storefront 1 and 1a) involved a 15-day return window. According to trade publications, this return condition is more conservative than retail-industry averages. Storefront2 and 2a received a 30-day return window, which corresponds closely with retail-industry averages. Finally, Storefront3 and 3a received a 60-day return window, which is more liberal than many retailers offer at this point. The other elements of the return remained constant across the three storefronts. Therefore, in terms of overall return-policy liberalness, it could be argued that Storefront1/1a < Storefront 2/2a < Storefront 3/3a . It is important to note that it is very common in my data that we observe a customer’s bidding behavior in several auctions. [Question] During the revision stage of the journal publication process, one of reviewer’s comment was that the I may use a fixed effects model to control for unobserved individual fixed effects. Do you agree or disagree with the above statement? Please explain with details.

Answer 1

Yes, the statement is correct. In case of panel data estimation, in order to control for the unobserved individual fixed effects, it is recommended to use fixed effect model. In most of the panel data models, we use fixed effect model to control for unobserved fixed effects in the model. This is because fixed effect model is the model in which model parameters are fixed and are non random quantities.The levels of the independent variable are fixed and control for unobserved fixed effect in the model.