DATA Step, Macro, Functions and more

PROC MIXED

Reply
Established User
Posts: 1

PROC MIXED

[ Edited ]

I am using SAS PROC MIXED. Could you please give your comments? It would be easier to read the attached file.

 

SAS PROC MIXED:

Sales for year t of firm i in industry j (sales_(tij)) as a function of an intercept (beta_{0}) plus a residual or error term (xi_{tij}).

 

sales_{tij} = beta_{0}+ xi_{tij}

xi_{tij} = eta_{j}+nu_{i(j)} + epsilon_{t(ij)}

 

rho_{z}=sigma_{z}^2/(sigma_{eta}^2+sigma_{nu}^2+sigma_{epsilon}^{2), z = eta, nu, epsilon

 

I use SAS PROC MIXED and SAS code:

 

PROC MIXED DATA=WORK.IMPORT METHOD=REML NOCLPRINT COVTEST;

CLASS industry id;

model sales = /solution ddfm=bw;

random intercept / subject=industry;

random intercept / subject=id(industry);

RUN;

 

I find rho_epsilon=0.22, rho_nu=0.62, and rho_eta=0.16. Now I want to ask the following two questions-

(1) How much this among firms within industries variation (rho_nu=0.62) is explained by firm age and size?

(2) How much within firms over time variation (rho_epsilon=0.22) is explained by recessions?

 

To address these two questions, how could I write the SAS PROC MIXED code? Is there any method do you suggest to address these two questions?

 

Thanks for your comments.

 

Ask a Question
Discussion stats
  • 0 replies
  • 292 views
  • 0 likes
  • 1 in conversation