It’s certainly not ideal from a pairings standpoint, but in some cases where the event has been advertised as 5 rounds and one of the sections turns out to have only 8 or 10 players, it’s better to let them play and pair it as fairly as possible. For scholastic events, you can sometimes combine sections, but that’s not always a good fit. As you point out, you can decrease the number of rounds - which means you can afford to lengthen the time controls. But younger players are not likely to use the extra time anyway. If the only objective is to determine a winner, then, yes, play 3 rounds and go home. But in my experience, most scholastic players and parents who have made the effort to get to the tournament prefer to play the scheduled 5 rounds, even knowing in advance that the pairings will get ugly. Everybody has a full day of chess, and goes home happy. 
If I’ve come to play in a four round Swiss, I don’t want it reduced to 3 rounds just because there are few enough players to get a clear winner after 3 rounds. Use the modified round robin format, or take your chances that you can get the tournament to pair right.
I think you have to give people what you advertised, and play however many rounds noted.
Occassionally in a 4-round tournament with farily short time controls (eg G/45), I’ve had only 4-6 or so show up, and gave them the choice of dropping one round and adding more time, with the proviso it had to be unanimous to make the change. Generally, my experience has been local players like to add time, but out of towners want to play the number of games they made the trip for.
Grant