Recoding was not carried out.
The major problem is that the 1992 dataset has codes for started education,
 whereas in 2000 only completed education was collected.
Also, the codes are mainly based on years in education, not highest attained level.
LIS made an attempt to recode started or tried education to the next lower level,
 but the frequency distribution does not correspond to the 2000 dataset.

If LIS users want to pursue a recode they may do so on their own responsability.
The mapping below may serve as suggestion.
PEDUC							     ISCED97	   LIS recode
  0  no formal education							1
  1  1 year school								1
  2  2 years school								1
  3  3 years school								1
  4  4 years school								1
  5  5 years school								1
  6  6 years school								1
  7  7 years school								1
  8  8 years school								1
  9  9 years school								1
 10  10 years school						3A		2
 11  11 years school						3A		2
 12  0-6 years + tried vocational						1
 13  7-9 years + tried vocational						1
 14  10-11 years + tried vocational						2
 15  0-6 years + finished vocational						1 or 2 ?
 16  7-9 years + finished vocational						2
 17  10-11 years + finished vocational						2
 18  0-6 years + started tech							1
 19  7-9 years + started tech 							1
 20  10-11 years + started tech							2
 21  0-6 years + finished tech							2 
 22  7-9 years + finished tech							2 or 3 ?
 23  10-11 years + finished tech						3
 24  0-6 years + started college						1
 25  7-9 years + started college						1
 26  10-11 years + started college						2
 27  0-6 years + finished college						3
 28  7-9 years + finished college						3
 29  10-11 years + finished college						3
 31  7-9 years + some postgraduate						3
 32  10-11 years + some postgraduate						3
 35  10-11 years + finished graduate degree					3

REMARK about recoding :
Source : UNESCO/ World Education Forum