multiple imputation with missing outcome data longitudinal study

[katy-barry]

Hello, 

  I am working with a dataset in which we measure if a person was unemployed in different years. for example, we ask "are you currently unemployed" in 2009, 2011, 2014, etc. I am wanting to do multiple imputation on my dataset but I do not know if I should add the outcome into my model because for some waves of data, people did not answer the question and if I include it into the multiple imputation model, SAS will generate values for my outcome variable which i do not want ( I know with multiple imputation, you just want the covariates to be filled in but not the outcome variable). Normally, your outcome and main exposure variable are already complete but since I have a longitudinal design, not all the outcome variables are completed. 

should I add the main outcome variable to the imputation model or leave it out?

thank you