[R] Plm function, error in uniqval[as.character(effect), , drop = F]

[lijuan huang]

Hi, there,

I am trying to run a panel regression with a huge dataset, which has lots
of missing values.  Here is some description of all variables.

> str(com3)
Classes ‘plm.dim’ and 'data.frame': 172153 obs. of  30 variables:
 $ tic                 : Factor w/ 3435 levels "A","AA","AAC",..: 1 1 1 1 1
1 1 1 1 1 ...
 $ idate               : Factor w/ 81309 levels "2001/12/31 0:27",..: 34
2371 2450 4789 5977 7297 7650 9988 11288 12749 ...
 $ IS_EFFECTIVE        : int  1 1 1 1 1 1 1 1 1 1 ...
 $ MATERIAL_WEAKNESS   : int  0 0 0 0 0 0 0 0 0 0 ...
 $ SIG_DEFICIENCY      : int  0 0 0 0 0 0 0 0 0 0 ...
 $ NOTEFF_ACC_RULE     : int  0 0 0 0 0 0 0 0 0 0 ...
 $ NOTEFF_FIN_FRAUD    : int  0 0 0 0 0 0 0 0 0 0 ...
 $ NOTEFF_OTHER        : int  0 0 0 0 0 0 0 0 0 0 ...
 $ IC_OP_TYPE_a1m0     : int  NA NA NA NA NA NA NA NA NA NA ...
 $ AUDITOR_FKEY        : int  NA NA NA NA NA NA NA NA NA NA ...
 $ AUDITOR_AGREES      : int  NA NA NA NA NA NA NA NA NA NA ...
 $ COMBINED_IC_OP      : int  NA NA NA NA NA NA NA NA NA NA ...
 $ IC_IS_EFFECTIVE_Y1N0: Factor w/ 4 levels "","0","1","ND": NA NA NA NA NA
NA NA NA NA NA ...
 $ AUDIT_FEES          : int  NA NA NA NA NA NA NA NA NA NA ...
 $ NON_AUDIT_FEES      : int  NA NA NA NA NA NA NA NA NA NA ...
 $ BENEFITS_FEES       : int  NA NA NA NA NA NA NA NA NA NA ...
 $ IT_FEES             : int  NA NA NA NA NA NA NA NA NA NA ...
 $ TAX_FEES            : int  NA NA NA NA NA NA NA NA NA NA ...
 $ AUDIT_RELATED_FEES  : int  NA NA NA NA NA NA NA NA NA NA ...
 $ OTHER_FEES          : int  NA NA NA NA NA NA NA NA NA NA ...
 $ CARD                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ DISC                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ HACK                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ INSD                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ PHYS                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ PORT                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ STAT                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ UNKN                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ BRCH                : int  NA NA NA NA NA NA NA NA NA NA ...
 $ ddate               : Factor w/ 81306 levels "2001/12/31 0:27",..: 34
2371 2450 4789 5977 7297 7650 9988 11288 12749 ...

And I ran a panel regression as follow:

rm(list=ls())
library(plm)
como3 <- read.csv(file.path('como3.csv'),header = T, stringsAsFactors = F)
com3=plm.data(como3, indexes = NULL)
summary(com3)

Y <- cbind(com3$BRCH)
X <-
cbind(com3$IS_EFFECTIVE,com3$MATERIAL_WEAKNESS,com3$SIG_DEFICIENCY,com3$IC_IS_EFFECTIVE_Y1N0,com3$AUDITOR_AGREES,com3$COMBINED_IC_OP,com3$AUDIT_FEES)

fixed <- plm(Y ~ X,com3, index = c("tic","idate"),effect="individual",model
= "within")
summary(fixed)

Unfortunately, I got this error:

series IT_FEES, BRCH are constants and have been removedError in
uniqval[as.character(effect), , drop = F] :
  incorrect number of dimensions


BRCH is my dependent variable, which is a dichotomous variable (happened
=1, unhappen=0). I don't know why it is considered as the "constant"? Does
it cause this error?

I have been looking for a solution in the internet for several days, but
have not found an answer to fix the problem.

Does anybody have an idea what I am doing wrong and what must be done to fix
the problem?

Thanks in advance.

Lynn Huang

	[[alternative HTML version deleted]]