This thesis proposes a decentralized procedure for optimal allocation of assets in loan divisions within commercial banks; simulation runs of the procedure demonstrate the feasibility of such an establishment in commercial banks. The decentralized procedure can not only reduce information inputs of the asset allocation system, but also provide a systematic procedure for estimations of the cost of funds and the cost of risk used in the profitability analysis. The simulation results also suggest that the appropriate price paid for the future customer relationships should vary according to the level of loanable funds available to the bank. Moreover, the simulations reveal that the interaction of risk within a bank is very complicated and point out the need for establishing some sort of procedures to take into account these interactions. / Business, Sauder School of / Graduate