# treatment of non-consecutive periods in panel and ʀ SFA

ɪ am estimating a Stochastic Frontier Analysis (SFA estimated by MʟE) model for a panel of seven periods. ɪ want to know if ɪ can use all the periods in the estimation procedure and how are they treated.

year
2005
2006
2008
2010
2011
2014
2016


ɪn case it is possible, ɪ used ʀ pdata.frame to declare de panel dataː

complete3 <- pdata.frame(mydata,index=c("country","year"))
str(complete3\$year)
Factor w/ 12 levels "2005","2006",..: 1 2 4 6 7 10 12 1 2 4 ...


ɪ sued sfa, from the package frontier, to run the model.

trialtime<-sfa(mymodel,
data = complete3, ineffDecrease=FALSE,  timeEffect = TRUE)


how do ɪ check that the inefficiency errors follow the equation defined in the model for each yearʔː

$$U_{it}=U_i \cdot e^{-\eta(t-T)$$