restricting the sum of coefficients

junjee713 · Postby **junjee713** » Sun Feb 14, 2010 10:30 pm

how do i restrict the sum of coeffients to sum to 1? i am doing a GMM model on the new keynesian phillips curve and i want to restrict the sum of two coefficients. my model specification is: gdpd=c(1)+c(2)*gdpd(1)+c(3)*gdpd(-1)+realmc. i want c(2)+c(3)=1. i have read two related topics in this forum on this topic and i tried to follow the suggestions there (e.g. (1-c(1) or 1-exp(c(1)). instead of c(3), i use (1-c(2))*gdpd(-1). however, when i run my regression, the results do not report the coefficient and related info for gdpd(-1). i'd like to know what am i doing wrong?

startz · Postby **startz** » Mon Feb 15, 2010 8:24 am

If you know c(2), 1-c(2) isn't that hard to figure out manually. :)

junjee713 · Postby **junjee713** » Mon Feb 15, 2010 5:52 pm

yes that is true but i want to know the associated s.e., t-stat, p-value as well.

startz · Postby **startz** » Mon Feb 15, 2010 6:27 pm

The standard error of 1-c(2) is the same as the standard error of c(2).

junjee713 · Postby **junjee713** » Mon Feb 15, 2010 9:20 pm

oh ok, thanks a lot for the help. :D

i have another questions. i am also using a 2sls model and i want to do the newey west HAC. eviews has this but it doesn't have an option to set the number of lags. what if i want to do so? how do i do it? thanks again.

EViews Gareth · Postby **EViews Gareth** » Mon Feb 15, 2010 11:26 pm

EViews 7 lets you specify the number of lags for TSLS in single equations.

junjee713 · Postby **junjee713** » Tue Feb 16, 2010 2:52 am

i see but i am using eviews 6. is there a way to do it in eviews 6?

EViews Glenn · Postby **EViews Glenn** » Tue Feb 16, 2010 10:23 am

Estimate your single equation in a system.

junjee713 · Postby **junjee713** » Tue Feb 16, 2010 5:05 pm

k, will do. thanks!

junjee713 · Postby **junjee713** » Wed Feb 17, 2010 12:41 am

i tried estimating my equation in a system as suggested:

(b=12) gdpd=c(1)+c(2)*gdpd(1)+c(3)*gdpd(-1)+lrealmc
@instr gdpd(-1 to-4) ygap(-1 to-2) nw inf lrmc

with b=12, number lags of newey west. it doesn't seem to work. is my specification wrong? :-(

EViews Glenn · Postby **EViews Glenn** » Wed Feb 17, 2010 12:10 pm

Yes, it's wrong. You'll have to estimate using GMM with the TSLS weighting matrix. See the discussion of System GMM.

prith · Postby **prith** » Thu Dec 16, 2010 5:29 am

Hi..same question..how do I restrict the sum of 2 coefficients to equal 1 in the equation.

inflm=c(1)+c(2)*ipgap+c(3)*infl_e+c(4)*inflm_l

Would like inflm_l+infl_e =1

EViews Gareth · Postby **EViews Gareth** » Thu Dec 16, 2010 8:21 am

inflm=c(1)+c(2)*ipgap+c(3)*infl_e+(1-c(3))*inflm_l

EViews.com