Estimating Corporate Yield Curves

Tests of arbitrage-free pricing models show that we can expect less reliable corporate yield curve estimates than Treasury yield curve estimates. An examination of the structure of errors produced by common statistical yield curve models indicates that even with careful data selection, significant liquidity and tax-induced errors remain. It is a welcome result that the extent of the errors due to liquidity and tax effects is modest. Moreover, pooling bonds by broad rating category produces no significant deterioration in yield curve estimates.