Dear Members,
I would like to run a two-stage hedonic model to establish whether consumers of a number of food products are willing to pay (WTP) a premium for country of origin declared on the label
My data contains a monthly panel for average unit (kg) sales price, quantities sold, country origin for each product and a set other covariaties which may capture attributes of meat from different countries.
I am using a product unique code (i.e. supermarket barcode) panel variable or id, and monthly records for the sales as a time variable(t). The goal is estimate (a) how much are consumers are WTP for a particular barcode of a country origin, and (b) what are key factors that influence WTP across products or fixed (SKU-specific) effects.
At the first stage of the hedonic model I would like to estimate the implicit price paid by consumers for countries of origin by estimating equation pijt= (αi+cj)+βxijt+ θZ + eijt, where αiis an overall intercept and c is time-invariant fixed effects for each product i. Z is a vector of all covariates. For more details on the specification, see Schulz et al. (2012), Value of Beef Steak Branding : Hedonic Analysis of Retail Scanner Data, Agricultural and Resource Economics Review, 41/2 (August) 260–273.
At the second stage, I intend to use the estimated implicit price paid by buyers for the country of origin of product i by regressing β on Z covariates. The goal is to estimate how match of WTP for country of origin is determined by other attributes.
I have two questions:
Question 1: how I can write β as the dependent variable for the 2nd stage in the Stata command line?
Question 2: Would it be appropriate to estimate two stages separately or use -xtsur to estimate the two stages jointly. And if yes, how β from first stage has to be writen in the command line?
Thanks,
Mohamud
I would like to run a two-stage hedonic model to establish whether consumers of a number of food products are willing to pay (WTP) a premium for country of origin declared on the label
My data contains a monthly panel for average unit (kg) sales price, quantities sold, country origin for each product and a set other covariaties which may capture attributes of meat from different countries.
I am using a product unique code (i.e. supermarket barcode) panel variable or id, and monthly records for the sales as a time variable(t). The goal is estimate (a) how much are consumers are WTP for a particular barcode of a country origin, and (b) what are key factors that influence WTP across products or fixed (SKU-specific) effects.
At the first stage of the hedonic model I would like to estimate the implicit price paid by consumers for countries of origin by estimating equation pijt= (αi+cj)+βxijt+ θZ + eijt, where αiis an overall intercept and c is time-invariant fixed effects for each product i. Z is a vector of all covariates. For more details on the specification, see Schulz et al. (2012), Value of Beef Steak Branding : Hedonic Analysis of Retail Scanner Data, Agricultural and Resource Economics Review, 41/2 (August) 260–273.
At the second stage, I intend to use the estimated implicit price paid by buyers for the country of origin of product i by regressing β on Z covariates. The goal is to estimate how match of WTP for country of origin is determined by other attributes.
I have two questions:
Question 1: how I can write β as the dependent variable for the 2nd stage in the Stata command line?
Question 2: Would it be appropriate to estimate two stages separately or use -xtsur to estimate the two stages jointly. And if yes, how β from first stage has to be writen in the command line?
Thanks,
Mohamud
