On the use of capacities in representing premium calculation principles