chrisconlon
commented
1 year ago Yes, you can use some marginal cost shifter such as wholesale price in two places:
- As a demandinstrument z{jt}^d
- In the supply side since you are specifying mc_jt = wjt * gamma + \omega{jt} in the X3 formula.
The main extra benefit from specifying supply in addition to just including a strong cost shifter in z_{jt}^d is that you are also specifying the conduct of the price setting firms. That is you are imposing the constraint MR = MC once you specify a supply side (and X3 formula). You can either specify the firm_ids or a full ownership matrix. Obviously this is a strong (but powerful) assumption, and it is stronger than simply saying "here is something correlated with prices but not the demand shock".
-Chris
On Thu, Dec 29, 2022 at 2:54 PM mtvseven @.***> wrote:
I have a question that is more theoretical than programming-related regarding the implementation of the supply side.
The data I have is transaction-level data on both products being sold and some information on customers from an online retailer. In order to familiarize myself with the mechanics of your package, I estimated the plain logit, am now adding a supply side without random coefficients, and will add random coefficients in the near future.
I have full visibility into cost data, so for plain logit I used wholesale price (one of the firms input prices) as an excluded instrument for price. For supply I am wondering if it makes theoretical sense to include wholesale price as both an excluded demand_instruments and as a cost shifter in X3 or if you know of any theoretical issues this would cause even if the optimization routine converges without issues?
Thanks!
— Reply to this email directly, view it on GitHub https://github.com/jeffgortmaker/pyblp/issues/124, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA7IOWOVQMGBIE6ATIFH5DTWPXT7FANCNFSM6AAAAAATMLHJNI . You are receiving this because you are subscribed to this thread.Message ID: @.***>
mtvseven
I have a question that is more theoretical than programming-related regarding the implementation of the supply side.
The data I have is transaction-level data on both products being sold and some information on customers from an online retailer. In order to familiarize myself with the mechanics of your package, I estimated the plain logit, am now adding a supply side without random coefficients, and will add random coefficients in the near future.
I have full visibility into cost data, so for plain logit I used wholesale price (one of the firm's input prices) as an excluded instrument for price. For supply I am wondering if it makes theoretical sense to include wholesale price as both an excluded demand_instruments and as a cost shifter in X3 or if you know of any theoretical issues this would cause even if the optimization routine converges without issues?
Thanks!