I heard an analogy between writing without a framework / your own framework an nuclear energy that I found fitting (not taking sides either way). Please bear with me.
The best-case scenario is better. More energy in the case of nuclear and just what you need, more performant, less code in the case of no framework. Everything just for your use case and no more.
But the worse-case is much much worse too. If you drop checks and discipline once (key developer switching jobs without in-team knowledge, for instance) there is no way back and the consequences can be catastrophic.
I would conclude than in general it is not worth it, but if you have the discipline and political incentives and/or special requirements (big ifs) it can be worth it, as long as those ifs keep being true.
The best-case scenario is better. More energy in the case of nuclear and just what you need, more performant, less code in the case of no framework. Everything just for your use case and no more.
But the worse-case is much much worse too. If you drop checks and discipline once (key developer switching jobs without in-team knowledge, for instance) there is no way back and the consequences can be catastrophic.
I would conclude than in general it is not worth it, but if you have the discipline and political incentives and/or special requirements (big ifs) it can be worth it, as long as those ifs keep being true.
I guess my insight is: it depends :)