topic Re: Panel regression with fixed effects and clustered standard errors in Statistical Procedures

Panel regression with fixed effects and clustered standard errors

reza96 — Tue, 04 Jul 2017 14:45:50 GMT

Hello,

I am using the SAS 9.4 edition and I would like to conduct the following panel regression : trade_{i,j,t} = Intercept + B1 * flow_{i,t} + …. Epsilon_{i,t}. The dependent variable represents the percentage change in stock holding of fund i in stock j at quarter-end t relative to time t-1, flow_{i,t} is the capital flow in/out of fund i at quarter-end t.

Which procedure do I need to use to estimate this regression which includes fixed effects at each quarter-end t and which clusters standard errors at the fund level i?

Below is the code I am using now, which only clusters standard errors at the fund level. Thanks in advance for your help.

proc surveyreg data= partial_scaling;

cluster fund;

weight regression_weight;

model perc_delta_shares = next_qpercflow / ADJRSQ solution ;

quit;

Regards,

Reza

Re: Panel regression with fixed effects and clustered standard errors

Sinistrum — Fri, 02 Feb 2018 14:53:24 GMT

Hello,

although presumably you will no longer be in need of an answer, as search machines might redirect someone considering the same issue (well, at least me) to this place:

proc	surveyreg
	data					=	partial_scaling;
	cluster						fund;
	class						quarter_end;
	weight						regression_weight;
	model
	perc_delta_shares		=	next_qperfclow	quarter_end
								/
								ADJRSQ
								solution;
quit;

For a large number of fixed effects, I suggest this macro.

Yours sincerely,

Sinistrum

Re: Panel regression with fixed effects and clustered standard errors

sasphd — Tue, 18 Sep 2018 18:16:16 GMT

what de you mean by

regression_weight

Re: Panel regression with fixed effects and clustered standard errors

Saba1 — Tue, 25 Jun 2019 05:11:55 GMT

@reza96 I have a similar requirement for my regression analysis i.e., clustered S.E. at fund level and year fixed effect. Please guide me which code has helped you achieving this objective. Thanks.

What is the purpose of "regression_Weight"?