Refunding Efficiency: A Generalized Approach

Refunding efficiency, a measure of the optimality of a refunding decision, is widely used in the call exercise decision for agency, corporate, and municipal bonds. The original definition of efficiency assumes that the refunding bond is optionless. However, in practice, the refunding bond is often callable. We show that the commonly used method of incorporating the value of the refunding bond’s call option into the efficiency calculation can lead to paradoxical results, and then suggest a new definition of refunding efficiency that overcomes this problem. Refunding Efficiency: A Generalized Approach Introduction Call provisions are a common feature of bonds issued in the United States. Most of the roughly 15,000 agency bonds outstanding are callable, as are most municipal bonds whose maturity exceeds 10 years. Until recently the majority of long-term bonds issued by electric and telephone utilities also included call provisions. The total volume of callable bonds outstanding is several trillion dollars. In recent years as interest rates declined the call/refunding volume in the Agency and municipal sectors has been running at a pace of hundreds of billions of dollars annually. This volume includes the so-called advance refunding (defeasance) of municipal bonds, which is effectively a call exercise prior to the initial call date (Kalotay and May, 1998). The associated transaction cost has been on the order of billions of dollars per year. Clearly, bond refunding is big business, and issuers of callable bonds are inundated by investment banking proposals recommending related transactions. Many of these proposals routinely provide the so-called refunding efficiency of the transaction. According to standard option pricing theory, a bond should not be called until the savings achieved by refunding equals the value of the call option. However, risk-averse issuers may not want to wait until the absolute optimal moment and will sometimes call the bond prematurely. A measure of how close a premature call is to being optimal is therefore useful.