[R-sig-Geo] Estimate impacts from a spgm - spatial panel regression model

Jose Funes je|une@ @end|ng |rom gm@||@com
Sat May 27 16:58:04 CEST 2023


Thanks Professor Bivand and Josiah. See below, basic info and code:

# Rstudio version and PC info

platform       x86_64-w64-mingw32
arch           x86_64
os             mingw32
system         x86_64, mingw32
status
major          4
minor          1.2
year           2021
month          11
day            01
svn rev        81115
language       R
version.string R version 4.1.2 (2021-11-01)



####### code #####

# Load necessary libraries
library(tigris)
library(splm)
library(spdep)
library(Matrix)
library(plm)

# census tracts
dc_tracts_2018 <- tracts(state="DC",cb=TRUE, year=2018)
dc_tracts_2018$row.id <- 1:179

# Spatial matrix
coords <- coordinates(dc_tracts_2018)
DC.knn_7 <- knearneigh(coords, k=7)
DC7_nb <- knn2nb(DC.knn_7)
DCknn_7 <- nb2listw(DC7_nb, style="W")

# reproducible example
set.seed(123)
df2010 <- data.frame(
  y = rnorm(179),
  x = rnorm(179, mean=2),
  x2 = rnorm(179, mean = 3),
  year=2010,
  row.id = 1:179
)

df2015 <- data.frame(
  y = rnorm(179),
  x = rnorm(179, mean=2.5),
  x2 = rnorm(179, mean = 3),
  year=2015,
  row.id = 1:179
)

df2020 <- data.frame(
  y = rnorm(179),
  x = rnorm(179, mean=1.5),
  x2 = rnorm(179, mean = 3.5),
  year=2020,
  row.id = 1:179
)


# drop geom
st_geometry(dc_tracts_2018) <- NULL

#list of data frame
df <- list(df2010,df2015,df2020)

for (i in seq_along(df)) {
  df[[i]] <- merge(df[[i]], dc_tracts_2018, by = "row.id")

}

df2010 <- df[1]
df2015 <- df[2]
df2020 <- df[3]

# merging
panel_dc_2010_2020 <- rbind(df2010[[1]], df2015[[1]], df2020[[1]])

#### setting lag variables
panel_dc_2010_2020.pd<-pdata.frame(panel_dc_2010_2020, index=c("GEOID","year"))

# lag variables
panel_dc_2010_2020.pd$lag_x <- slag(panel_dc_2010_2020.pd$x,listw=DCknn_7)
panel_dc_2010_2020.pd$lag_x2 <- slag(panel_dc_2010_2020.pd$x2,listw=DCknn_7)

## impacts
time1 <- length(unique(panel_dc_2010_2020$year))
s.lwtracts <- kronecker(Diagonal(time), listw2dgCMatrix(DCknn_7))
trMatc <- trW(s.lwtracts, type = "mult")

Error in trW(s.lwtracts, type = "mult") : could not find function "trW"

# durbin
model_fix_durbin <- spgm(y ~ x + x2 +lag_x + lag_x2,
data=panel_dc_2010_2020.pd, listw = DCknn_7, model = "within",lag =
TRUE,spatial.error = FALSE,moments="fullweights",method="ec2sls")

# estimating impacts
impacts(model_fix_durbin,tr=trMatc, R=200)

Error in impacts(model_fix_durbin, tr = a, R = 200) :
  could not find function "impacts"




On Fri, May 26, 2023 at 2:34 PM Roger Bivand <Roger.Bivand using nhh.no> wrote:

> I think the example is using the splm package to fit a GMM model. It is
> now some years since spdep was split, with model fitting functions and
> methods moving to spatialreg. The latter has also been updated several
> times since then, so knowledge of versions of R and packages is essential,
> in addition to a reproducible example. Specifically,
> https://github.com/r-spatial/spatialreg/pull/16 changed impacts methods
> for sphet GMM, and other changes in impacts methods may have occurred:
> https://github.com/r-spatial/spatialreg/commits/main/R/impacts.R gives
> full details. Provide a fully reproducible example, best with built-in
> data, or raise an issue if you can see how any changes have impacted splm
> in ways that the splm maintainer did not anticipate.
>
> Hope this clarifies,
>
> Roger
>
> --
> Roger Bivand
> Emeritus Professor
> Norwegian School of Economics
> Postboks 3490 Ytre Sandviken, 5045 Bergen, Norway
> Roger.Bivand using nhh.no
>
> ________________________________________
> From: R-sig-Geo <r-sig-geo-bounces using r-project.org> on behalf of Josiah
> Parry <josiah.parry using gmail.com>
> Sent: 26 May 2023 20:14
> To: Jose Funes
> Cc: r-sig-geo using r-project.org
> Subject: Re: [R-sig-Geo]  Estimate impacts from a spgm - spatial panel
> regression model
>
> Jose have you built the model that you want impacts for using the
> spatialreg package? Additionally, it would be great if you provided a
> reproducible example for folks to try and replicate the behavior you are
> experiencing.
>
> On Fri, May 26, 2023 at 12:26 PM Jose Funes <jefunes using gmail.com> wrote:
>
> > Hi,
> > I would like to estimate the impacts of a spatial lag model output. It
> used
> > to work, but it does not work anymore. I used to work fine with the
> impacts
> > function in the spdep package, but it seems the function does not work
> any
> > longer. The second option was to load the spatialreg library and use the
> > impacts function, but I get an error message "object type not
> recognized".
> > Please advise.
> >
> > Jose Funes
> > Economic geographer
> > DC Office of Planning
> > 1100 4th Street SW, 20024
> >
> >         [[alternative HTML version deleted]]
> >
> > _______________________________________________
> > R-sig-Geo mailing list
> > R-sig-Geo using r-project.org
> > https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> >
>
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

	[[alternative HTML version deleted]]



More information about the R-sig-Geo mailing list