R - Hausman Test

I'm trying to do a Hausman Test in order to compare the efficiency between Fixed Effects and Random Effects, but I'm receiving the following error:

Error in solve.default(dvcov) : 'a' is 0-diml.

I searched and found that this error could be because of covariance matrix is ​​not a invertible. I want to know if there is a solution for this or if I will need to search for another test.

Code:

fixed <- plm(Y ~ X, data=pCoopCred, model="within")

random <- plm(Ymil ~ Xmil, data=pCoopCred, model="random")

phtest(random,fixed)

p.s.: for Random Effects I had to divide the values ​​by 1000 because of variable lengths. Should I did this to Fixed Effects too?

Thanks !