Forum: help

RE: Pooled SUR with no intercept -> "(Intercept) = 0" is not well formed.... [ Reply ]
By: Arne Henningsen on 2019-03-28 09:04

[forum:46635]

Dear M.P.

I have insufficient information about your estimation to give you a precise answer but I will give you some hints. If you want to suppress the intercept in one or more equations, you can add "- 1" to the model formula of these equations. Your restriction causes an error message, because there is no coefficient with name "(Intercept)". The system of equations usually has multiple intercepts (one per equation). In order to clearly indicate which coefficient belongs to which equation, the name of each coefficient includes the name of the corresponding equation, e.g. "myFirstEquation_(Intercept)". The documentation of systemfit() says: "The [...] names of the coefficients can be obtained by estimating the system without restrictions and applying the coef method to the returned object." If you have further questions, please provide a minimal reproducible example.

Best regards,
Arne

Pooled SUR with no intercept -> "(Intercept) = 0" is not well formed.... [ Reply ]
By: M. P. on 2018-12-26 17:45

[forum:46478]

Hi,

I have panel data, from which the time and individual fixed effects were subtracted (see last equation in: https://stats.stackexchange.com/questions/246548/difference-between-one-way-and-two-way-fixed-effects-and-their-estimation). I am doing a pooled SUR estimation of the coefficients and am now trying to get rid of the intercept term. I have tried to do the following:

m <- c("(Intercept) = 0")

model_fit <- systemfit(formula = EQUAT, method = "SUR", data = DF_pannel, pooled = TRUE, restrict.matrix = m)

but, I get the error message: "(Intercept) = 0" is not well formed: contains bad coefficient/variable names. Any ideas how to fix this and eliminate the intercept?