PLM in R with time invariant variable -
i trying analyze panel data includes observations each state collected across 45 years. have 2 predictor variables vary across time (a,b) , 1 not vary (c). interested in knowing effect of c on dependent variable y, while controlling , b, , differences across states , time.
this model have, using plm package in r.
random <- plm(y~log1p(a)+b+c, index=c("state","year"),model="random",data=data)
my reasoning time invariant variable should using random rather fixed effect model. question is: model , thinking correct?
thank in advance.
you base answer decision between fixed , random effect soley on computational grounds. please see specific assumptions associated different models. hausman test used discriminate between fixed , random effects model, should not taken definite answer (any textbook have further details).
also pooled ols yield model, if applies. computationally, pooled ols give estimates time-invariant variables.
Comments
Post a Comment