Hello together 
We want to run a stochastic frontier analysis to derive the efficiency of marketing capability on sales. The literature suggests different input factors for this purpose , for example:
And here comes the question: How can we derive the variable MARKETINGSTOCK using a Koyck lag structure? We want to use an finite approach using 5 years data. (we have paneldata for firms & years)
I know, that the underlying equation looks like this: Yt=a +ß0Xt + ß1Xt-1+ ... ßkXt-k + et
and that the Koyck approach puts greater weight on the factor with a lower distance in time.
I could find some example using regression models regressing lagged independent variables on a dependent variable. But as the variable MARKETINGSTOCK consists of the SGAEXPENSE over time, this would not be suitable for our case.
Does anybody know how to implement a code for getting what the screenshot says?
Or do we have to regress the lagged values of SGAEXPENSE on sales in order to get the coefficients and than use these coefficients for each lagged SGAEXPENSE to calculate the MARKETINGSTOCK?
We want to run a stochastic frontier analysis to derive the efficiency of marketing capability on sales. The literature suggests different input factors for this purpose , for example:
And here comes the question: How can we derive the variable MARKETINGSTOCK using a Koyck lag structure? We want to use an finite approach using 5 years data. (we have paneldata for firms & years)
I know, that the underlying equation looks like this: Yt=a +ß0Xt + ß1Xt-1+ ... ßkXt-k + et
and that the Koyck approach puts greater weight on the factor with a lower distance in time.
I could find some example using regression models regressing lagged independent variables on a dependent variable. But as the variable MARKETINGSTOCK consists of the SGAEXPENSE over time, this would not be suitable for our case.
Does anybody know how to implement a code for getting what the screenshot says?
Or do we have to regress the lagged values of SGAEXPENSE on sales in order to get the coefficients and than use these coefficients for each lagged SGAEXPENSE to calculate the MARKETINGSTOCK?
