This thesis considers the problem of allocating an available budget across investments proposed by several departments of an organization. While solely investments with positive returns are relevant, not only the total profit generated by the allocation is taken into account, but also the fairness of the monetary distribution across said organizational units. Therefore, profit and distributional fairness are the two important characteristics of a specific allocation in this work. Then, the aim was to find Pareto-efficient solutions in regard to the two competing objectives of profit and fairness for the underlying binary knapsack problem. Calculations were performed for both extreme points and several distinct minimal fairness requirements along the Pareto frontiers for a large number of problem instances. Afterwards, the solution sets across instances of each specific point of the frontiers were aggregated, yielding an estimation of the profit/fairness trade-offs behaviour between the two extremes. This procedure was performed for five different fairness criteria, where in theory each of them evaluates the fairness of the allocation differently. After the presentation of computational models used to calculate the mentioned Pareto-efficient solutions, extreme point solutions were presented for each criterion and compared among each other. Subsequently, the behaviour of the trade-off curves was displayed by the aggregated Pareto frontiers, showing the estimated loss of profit when considering higher levels of distributional fairness.Based on the achieved results, it can be said that in general the magnitude of sacrificed profit between profit maximization and increased fairness requirements appears rather small. Additionally, the shapes of estimated fairness criterion specific frontiers seem to be similar and lead to the assumption that the trade-off between profit and fairness follows a rather linear function. |