# Why do ARIMAX standard error differ in EViews vs. R while coefficients coincide?

The estimates by function Arima from the "forecast" package in R are as follows:

ARIMA(1,0,0) with non-zero mean

Coefficients:

ar1   intercept  SEASONAL.. 1MO_LIBOR... GDP_GOODS... CORP...

0.3950   0.0464    -0.0783    -0.0220       1.8730       0.0679

s.e.  0.1463   0.0068     0.0083     0.0115       0.8527       0.0323


Meanwhile, the EViews estimates are

Why do the standard error differ in EViews vs. R even though the coefficients coincide?

By the way, could any body help me to calculate the $p$-values of the coefficients? Is this function correct?

P_Value <- function(fit)
{
if(inherits(fit,"lm")){
res=summary(fit)$coef[-1,4] }else { vars=colnames(fit$xreg)
n=nobs(fit)
df=nobs(fit)-length(fit$coef) res=(1-pt(abs(fit$coef)/sqrt(diag(fit\$var.coef)*n/df),df))*2
}
return (res[vars])
}