The years and months of interview, birth of the respondent and her children, etc. are denoted in the ETHIOPIAN CALENDAR (7 to 8 years behind the Gregorian calendar). See page 10 of Individual Recode Documentation (etir41fl.doc zipped in etir41fl.zip). Although there are 13 months in the Ethiopian calendar, however, the month variables contain values from 1 to 12 only. I suppose they enter 12 for the 13th month.
The conversion to the Gregorian calendar may be available as country specific variables (those whose name begins with letter S). But I found out that the conversion messed up for a handful of observations. The following algorithm helps you convert years and months on your own:
If the month is 4 or less, add 7 to the year and add 8 to the month. Otherwise, add 8 to the year and subtract 4 from the month.
In the 2000 survey, the birth index and the birth order for country specific variables are messed up so that the lower indexed child is born earlier (it should be the opposite). Make sure you clean the data if you use country specific variables for children in your analysis.