A Specialized Inventory Problem in Banks: Optimizing Retail Sweeps

Deposits held at Federal Reserve Banks are an essential input to the business activity of most depository institutions in the United States. Managing these deposits is an important and complex inventory problem for two reasons. First, Federal Reserve regulations require that depository institutions hold certain amounts of such deposits at the Federal Reserve Banks to satisfy statutory reserve requirements against customers' transaction accounts (demand deposits and other checkable deposits). Second, some inventory of such deposits is essential for banks to operate one of their core lines of business: furnishing payment services to households and firms. Because the Federal Reserve does not pay interest on such deposits used to satisfy statutory reserve requirements, banks seek to minimize their inventory of such deposits. In 1994, the banking industry introduced a new inventory management tool for such deposits, the retail deposit sweep program, which avoids the statutory requirement by reclassifying transaction deposits as savings deposits. This is an interesting inventory problem for fungible items, where the conversion process is reversible. We examine two methods for operating such sweeps programs within the limits of Federal Reserve regulations, and we develop a stochastic dynamic programming model to implement one such method, the threshold method.