There are no livelihood descriptions for 93 and 98. If there were no GPS points for survey years 1993 and 98, is it possible to get these descriptions? Is there any other way to code the survey clusters?
Making available contextual variables on livelihood zones requires that two types of data are available: the data on livelihood zones from FEWSNET and GPS cluster data from DHS surveys. My guess is that the FEWSNET data on livelihood zones does not go back as far as you would like.
You can check the website for FEWSNET here: http://fews.net/fews-data/335
You don’t state which countries are of interest to you, but I can say that we did calculate contextual variables for all samples available at the time with public cluster data, and include them in the downloadable CSV files, available here: https://www.idhsdata.org/idhs/context…
The only samples not included are those released after we began calculating the contextual variables, and updated versions with the new samples will be released in the fall.
You might try mapping the livelihood zones available for more recent years onto provinces for countries of interest, and try to use DHS regions as a much less precise proxy for livelihood area. That might work passably for countries without a lot of variation, but would work less well for countries with a lot of internal variation, such as Zimbabwe. It seems likely that there is not a huge amount of change in livelihood zones over a decade, unless the area is very vulnerable to climate change and factors such as flooding or desertification.
Good luck. If you find a solution, please do share it in a post here.