When matching the PR data (based on the household schedule) with information from the KR file (based on children 0-59 months old in the woman's questionnaire) you can have 2 situations with missing data:

1) The child is in the PR file, but the mother was not interviewed and this child can thus not be found in the KR file.
2) The child 0-59 months is in the KR file, but has died or lives somewhere else and can thus not be found in the PR file.

You can exclude the children who have died or do not live in the household (b16 = 0) from the KR file before linking and this should eliminate 2).
But you should always have some children in the PR file without corresponding information in the KR file.

I hope this helps.