Because arrival rate can be tough to interpret
Most GDPs are issued 45 minutes prior to the first flight affected’s IGTD. While some are issued from status (flights about to push could be included), that usually indicates an unforcasted change in the AAR. As you have described, airborne holding and delay are two different things. At United, we found that airborne holding would reduce the actual accomplished AAR by 10-20%, as flights are not perfectly sequenced as they depart the holding patterns and fixes. I taught my ATC coordinators to advocate for other tactical TMIs such as MIT or APREQ for individual flights if a modeled GDP generated EDCTs of 15-18 minutes or less, and if the constraint on the AAR was a high value forecasted event. When some tactical delay is necessary, and a GDP is not used, it’s hard to pin down which flights might absorb the delay….although use of TBM (low visibility to the airline) release times might be sufficient. Good reasons why the CDM process is so valuable.
"But their wait-and-see approach betrays airlines' ability to plan..."
Story of my life- especially at a line station!