that is usually discussed under the term "branching storyline" - no matter if the branch is small or change the entire story. You can find a lot of discussions about it if you search it with that term instead of your "player agency".
That said it is something a lot of people and even more players would like, but it has two big problems:
1) workload
each branch of the story has to be developed and playtested, but usually the player only goes along one branch - they rarely start a game several times to follow all branches.
And that means the ratio between development time and effective playing time gets even worse - it is usually estimated that one playing hour needs 100 development hours, and if that playing hour is in reality just fifteen minutes of gameplay but in four different versions, then you're suddenly at 400 development hours for each playing hour of the player.
That is why professional games rarely do this unless they're specifically targeting something like that - and why a lot of promises of "player decisions count" do not really come to much change in professional games. Each development hour has to be paid for.
2) balancing
If there are many branches and they have different rewards, the developer can no longer know how strong a player is and which options and equipment he has at the end of the story, making the game balancing more difficult.