Actually, elevations of benchmarks have been changing for decades while we were still using the NGVD29 system. Each time Federal surveyors would come through Louisiana to re-observe benchmarks, a new epoch of benchmark elevations would become published by the National Geodetic Survey. With time, we discovered that the methodology was flawed because local benchmarks in the region were used as starting points, and their old elevations were used for the new surveys. Since those starting points had also subsided but were forced to remain having the same old elevations; each new survey provided different elevation values that were subsequently warped more and more.
With the new adjustment of the NAVD88, warps and the forcing of new surveys to fit old surveys are no longer done. The new elevations were compatible with each other when published in Louisiana in 1992, and although the original NAVD88 values are no longer valid (because of subsidence), the system of GPS CORS sites in Louisiana now allow elevation surveys to remain current as they are observed.