These "wage decoupling" studies generally make two methodological mistakes.
1) They do not include fringe benefits in their calculation of wages 2) They use one inflation measure to adjust for cost changes in productivity over time and a different inflation measure to adjust for cost changes in wages over time.
Doing a quick search over their methodology [1], to their credit, I believe EPI avoids mistake #1, but it appears they make mistake #2.
Studies which include the complete value of benefits in their calculation of wages and use the same method to deflate both wages and productivity show that there is no "wage decoupling" from productivity.