I think about that function with a probability parameter as well.
Let’s say that you have a probability of 73,88% of being sick for 2500 and of 26,12% of no medical bills this is how you would compare:
Deductible 300= 4450x73,88%+3930x26,12%=4314.176
Deductible 1500= 4732x73,88%+3132x26,12%=4314.08
Deductible 2500= 5017,2x73,88%+2467,2x26,12%=4351.14
So here is a mathematical valid case in with it’s worth taking an health insurance with a non-extreme deductible.
In reality that probability parameter is also uncertain. This means that you might estimate the right thing to take the 300 deductible but then you aren’t. Then next year you think that you will pay 3000 but then you pay 0. I that case it makes sense to go for a middle deductible.
To generalize:
- if you are bad at estimating how much you are going to spend it makes sense to go for a middle deductible.
- If you are extremly good at estimating, there are some rare cases (see above) in which it makes sense to go for a middle deductible.