How do I control for unit-specific trend in a panel regression with plm

Question

I am working with panel data using the plmpackage in R and need to specify a state and year fixed effects model with linear state-specific trends. I have the following code so far:

pnl_sb <- pdata.frame(seatbelt.df, index = c("fips","year"))
twyreg <- plm(fatalityrate~sb_useage+primary*sb_useage+secondary*sb_useage+speed65+speed70+ba08+drinkage21+log(income)+age, data = pnl_sb, model = "within", effect = "twoways")

fips is the unit variable states and year is the time variable. I simply do not know how to implement state-specific trends. I thought that it would be assumed with 2 way fixed effects panel data that added year as a time trend. But, I am having some doubts.

Maybe this help? https://stackoverflow.com/q/34232834/3834139 — Martin C. Arnold, Jan 09 '18 at 08:35
@M.A. thank you, I looked into this before I posted and did not find it helpful. I will be getting back to this project this afternoon and will look at it again then. — Collective Action, Jan 09 '18 at 08:54

score 0 · Answer 1 · answered Jan 10 '18 at 13:51

pnl_sb <- pdata.frame(seatbelt.df, index = c("fips","year"))
twyreg <- plm(fatalityrate~factor(state)*year, data = pnl_sb, model = "within", effect = "twoways")

Specifically: factor(state) ensures all of the states will be used. By multiplying this factor(state) by year trends are created for each state.

How do I control for unit-specific trend in a panel regression with plm

1 Answers1