Any observation in the input data set that has a missing value for the cross section ID, time series ID, dependent variable, or any model effect is ignored by PROC PANEL when it fits the model.
If your data contain observations in which only the dependent variable is missing, you can still compute predicted values for these observations and store them in an output data set by using the OUTPUT statement.