使用 plm 固定效果回归

Question

The code below runs the time series regression "excessr ~ mkt_rf" with biglm, because the function lm does not work with my real dataset.

Now I would like to switch from biglm to plm to account for fixed effects. Unfortunately plm doesn't work.

Does anyone know what I could change that plm work ? Thanks in advance!

library(biglm)
library(plm)
library(data.table)

union_with_factors = data.table(
  t = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5),
  excessr  = c(10,23,13,53,43,76,34,12,45,13,42,31,4,53,64),
  FundId = c("x","x","x","x","x","y","y","y","y","y","z","z","z","z","z"),
  mkt_rf = c(1,1,2,1,3,1,1,2,1,3,1,1,2,1,3)
)

sp <- split(union_with_factors, union_with_factors$FundId)
beta <- sapply(sp, function(tmp){
  fit <- plm(excessr ~ mkt_rf, data = tmp)
  coef(fit)
})

Answer 1

Supposing that the ID of individuals is given by FundId and that time ID is given by t , here's how you can apply a fixed effects regression:

library(biglm)
library(data.table)
library(plm)

union_with_factors = data.table(
  t = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5),
  excessr  = c(10,23,13,53,43,76,34,12,45,13,42,31,4,53,64),
  FundId = c("x","x","x","x","x","y","y","y","y","y","z","z","z","z","z"),
  mkt_rf = c(1,1,2,1,3,1,1,2,1,3,1,1,2,1,3)
)
fit <- plm(excessr ~ mkt_rf, 
             data = union_with_factors, 
             index = c("FundId", "t"), 
             model = "within")
summary(fit)
fixef(fit)

See here and in the plm documentation ( ?plm in console) for more details

Edit: following this post and this article , it appears that you can do a Fama-MacBeth regression with pmg (also in the plm package):

fama_macbeth <- pmg(excessr ~ mkt_rf, 
                   data = union_with_factors, 
                   index = c("FundId", "t"))
summary(fama_macbeth)