obtaining average coefficients and adj. R^2 from multiple pooled regressions using lapply

I have performed multiple Pooled regressions with a loop function and stored the regression output in a list (myregression). What I would like to do now is to efficiently perform the coeftest function in the lmtest package over all my regressions (i.e., myregression list) to adjust standard errors and t-statistics. Finally I would like to obtain the mean of the coefficients, standard errors and t-values.

Here is what I came up so far:

library(plm)
data("Grunfeld", package="plm")

# Store each subset regression in myregression
myregression <- list()

count <- 1

# Regression on six-year subsets of Grunfeld
for(t in 1940:1950){

  myregression[[count]] <- plm(inv ~ value + capital, 
                              subset(Grunfeld, year<=t & year>=t-5),
                              index=c("firm","year"))

# Name each regression based on the year range included in the data subset
names(myregression)[[count]] = paste0("Year_",t)
count <- count+1
}

Here is where my problems kick in: Although i'm able to perform the coeftest function to invidiual components of the list, I'm unable to code the lapply function accordingly.

## Apply coeftest function to all plm-objects
library(lmtest)

coeftest(myregression$Year_1940, vcov=function(x) vcovSCC(x, type="HC3", maxlag=4))
coeftest(myregression$Year_1941, vcov=function(x) vcovSCC(x, type="HC3", maxlag=4))

COEFTEST<-lapply(myregression, coeftest(x, vcov=function(x) vcovSCC(x, type="HC3", maxlag=4)))

## obtaining average coefficients, se's,and t values over all regressions
lapply(COEFTEST, mean)

I hope there is only a minor mistake that I'm unable to see. I further noticed that the plm regression output is smaller than regular lm output is there another way to obtain mean adj. R^2?

Solution

Try

 COEFTEST<-lapply(myregression, function(x) coeftest(x, vcov=vcovSCC(x, type="HC3", maxlag=4)))

Thsi will give you a list of coeftest-outputs for each regression, which you can then use in whatever way you want.

As a sidenote, make sure that whatever you do with this output makes sense. Taking the mean of an coeftest-output is not obvously sensible to me. If you want to have the average over all coefficients, try something like

 mean(sapply(COEFTEST, function(x)  x["capital", "Estimate"]))

Here, sapply retrieves all estimates for the variable capital from the COEFTEST outputs and puts them into a vector.

To access other elements, it is helpful to look at the structure of the objects using str(). For instance, str(summary(myregression[[1]])) reveals that the r-squares are saved under the name r.squared. These you can access using e.g. summary(myregression[[1]])$r.squared for the first regression output. Automating this, you can again use a construct as above, e.g.

 sapply(myregression, function(x) summary(x)$r.squared)