2. Off-peak trains have the exact same running costs as peak time trains. They usually have to be run with significant spare capacity to fulfill franchise agreements, and because it is not economic to chop carriages off and on to services in the middle of the day. To say that the extra cost is NIL is patently not true. And on a lot of the weekend (off-peak) trains, there can be significant overcrowding.
Without having exact numbers to hand I can see that there is a very strong, and universally accepted by economists, argument that the marginal cost of an extra off peak passenger is effectively nil. This is precisely because off-peak trains have the exact same running costs as peak time trains.
For example:
Lets say the route from Vicarstown to Tidmouth operates using 4 car trains with a capacity of 360. This stock has been selected because the demand at peak hours is around 300-340 per service.
At off peak hours the service has to be operated with the same stock as no other is available, however demand is around 100-120 per service.
The Marginal Cost of carrying 1 extra passenger on an off peak service is effectively zero, as unless the particular passenger generates extra work for staff on the service then the cost to the TOC of carrying that passenger is zero.
This means that (excluding the 100-120 people who usually use the service) a TOC could charge £1, and this would be pure profit, as costs have not increased. Assumption: 1 Guard is required who can check 360 tickets during the journey.
The problem in the case of Railcards is how do TOCs ensure that the passengers who travel on these off peak trains anyway do not result in the overall revenue declining. For example if you assume that 50% of the 100-120 passengers are travelling in pairs, and therefore make use of the new railcard, then Total Revenue would decline by 16.5%, so this new railcard (in this example) only results in an increase in revenue when overall passenger numbers increase by more than 16.5%.
So yes the marginal cost to the TOC of 1 extra off peak passenger is effectively zero. A revenue model has to be developed to support a pricing strategy such as a railcard, and the lack of a positive case for a National Railcard is the only reason I can see for it not being introduced.