Joint work with Peter Forsyth, University of Waterloo and Ken Vestal, University of Waterloo.

In recent years, core deposits such as checking accounts typically comprise about 35% of commercial bank liabilities. Valuing these liabilities has a large impact on bank balance sheets. This paper builds on previous work on demand deposit valuation. We value demand deposits based on data from RBC Finance Group, and determine the best numerical scheme under different scenarios. Numerical schemes such as Partial Differential Equation (PDE), Monte Carlo methods including crude Monte Carlo method and Antithetic Monte Carlo method, and Low discrepancy sequence methods (LDS) are tested. Convergence of different numerical schemes is compared. It is shown that the Monte Carlo method is adequate for short time horizons with low accuracy requirements, while the PDE method generates better results in other cases. Our result shows that these deposits are extremely profitable.

Keywords: Demand deposit, PDE, Monte Carlo, LDS