I have assumed that the shares will return 0.5% more than an ETF due to lower charges, but I might be wrong.
I would default to all having same growth performance. And model cost seperately.
If we are trying to model cost, I guess it depends on how good a model we are trying to build. or which lifecycle investment we are trying to model, e.g. lump-sum investor, accumulating investor, drawing down investor etc.,
I think there is potentially a few categories, not sure if we need them all though.
I think there is probably original investment/setup cost, onging investment maintenance cost, re-investment (or additional new investment) cost, drawdown costs
e.g. a buy and hold share investor, has some setup cost, has ~0 ongoing maintenance cost, had some cost to invest new money, and has some cost to drawdown from investment.
Alhough even then, if the dividend income > drawdown, then there would be no drawdown cost!
It might be simplest to assume we are focussed on the investor who has a lump-sum, or a lump-sum and accumulating more.