Hi,
I am using the package "pspatreg" to estimate all the models present in the following paper (An alternative semiparametric model for spatial panel data | SpringerLink).
I am using the "unrate" dataset that is linked to the package.
My question is about the intepretation of the coefficient in the estimation of the Non-Parametric models.
Once built the model, asking for the "summary", I can read some values (as in parametric regression) beside variables (indicated as "pspl(x)", "pspl" is used in case of non parametric variable). How can I interpret that values? Are them just mean values (so I have to consider only graphical analysis) or can I use them for some intepretation?
Here the code:
library(lmtest); library(car); library(dplyr);
library(sf); library(spatialreg); library(pspatreg);
library(plm); library(splm)
data("unemp_it", package = "pspatreg")
lwsp_it <- spdep::mat2listw(Wsp_it, style = "W")
#Formula to recall in the model
formnonlinpsanova <- unrate ~
pspl(empgrowth, nknots = 20) +
pspl(partrate, nknots = 20) +
pspl(agri, nknots = 20) +
pspl(cons, nknots = 20) +
pspl(serv, nknots = 20) +
pspt(long, lat, year,
nknots = c(18, 18, 12),
psanova = TRUE,
nest_sp1 = c(1, 2, 3),
nest_sp2 = c(1, 2, 3),
nest_time = c(1, 2, 3))
psanovanonlinsar <- pspatfit(formnonlinpsanova,
data = unemp_it,
listw = lwsp_it,
type = "sar")
summary(psanovanonlinsar)
Thanks a lot! 