just adding in another data point or two. I think it's a strong function of set-up as others have mentioned ...
My setup is: CT in the basement, temps around 12C up to 20C in summer, strong industrial main fan + 16" pedestal fan for 'trim' cooling. Old TV + VCR with lotsa old Tdf prologue, TT and mountain stage tapes (some back to the Indurain era!). Two windows close-by for ventilation (when it's not TOO cold outside). Good HRV unit on high to help keep humidity down. LOTS of water and some sugar-water close at hand.
It's a low-inertia setup but I mitigate that as much as possible by making sure I run the gearing/wheel-speed fairly high in ergo mode and by avoiding courses with steepish, sustained climbs (try to avg. from 35-45 kph). Pedalling thru mud every day isn't the most effective training IMHO!
So within those confines, over the years I've found most of the 'basement season' where temps are around 13-15C (summer is still an issue) power from a few minutes up to ~20min doesn't suffer at all compared to hard outdoor training. However, I still find some droop starts to occur around there that shows up, I reckon, as about 3% around the 30-min mark and around 5% at the full-hour mark.
Out past that, I believe the power loss is pretty constant - probably a function of the output power itself being sig. lower for say a 2-hr or 3-hr target session vs. hard threshold in the 30-60min region.
Now in summer when humidity is typically high and basement temps are 19-20-22C ... things are not copacetic. Generally speaking I'll stick to shorter efforts 20-30min long ... even <gasp> L5 work. Sustained hardish stuff just cooks me ...
All in all, I rate the power droop as mainly due to (still) inadequate cooling vs. good outdoor conditions. Then again, I suppose it hardly ever reaches 30C here so I'd have another complaint if it were that HOT!