I agree with your assumptions and your reasoning... however...
Do your calculations take into account the non-eTag collections?
If I were playing "Devil's Advocate", I'd say that the some of the higher "penalty" fee for non-eTag users (which won't be capped), will eventually be collected and added to the mix. But I can' even begin to suggest assumptions or estimations to even begin getting figures to plug into a formula.
The initial enforcement of non-eTaged motorists could scare the greater public and drive (excuse the pun) more people to get eTags.
Good job though... it does expose some gaping holes in their model.
However, Sanral will turn around and say they've followed models which have been tried and tested in other parts of the world.