What Determines U.S. Swap Spreads?
This paper examines the evolution of the U.S. interest swap market. The authors review the theory and past empirical studies on U.S. swap spreads, and estimate an error-correction model for maturities of 2, 5, and 10 years from 1994 to 2004. Financial theory depicts swaps as contracts indexed on London Inter-Bank Offered (LIBOR) rates, rendered almost free of counterparty default risk by mark-to-market and collateralization. Swap spreads reflect the LIBOR credit quality (credit component) and a liquidity convenience premium present in Treasury rates (liquidity component). Multifactor models that were estimated on observed swap rates highlighted the central role played by the liquidity component in explaining swap-spread dynamics over the past 15 years. The multifactor models also found some puzzling empirical results. Statistical models, on the other hand, based mainly on market analysis, faced technical difficulties arising from the presence of regime changes, from the non-stationary in swap spreads, and from the coexistence of long-term and shorter-term determinants. Against this background, the authors apply an error-correction methodology based on the concept of co-integration. They find that U.S. dollar swap spreads and the supply of U.S. Treasury bonds are co-integrated, suggesting that the Treasury supply is a key determinant on a long-term horizon. The authors estimate an error-correction model that integrates this long-term relationship with the influence of four shorter-term determinants: the AA spread, the repo rate, the difference between on-the-run and off-the-run yields, and the duration of mortgage-backed securities. The error-correction model fits observed swap spreads quite well over the sample period. The authors illustrate how the same model can be used to carry out scenario analysis.
Main Authors: | , , |
---|---|
Language: | English en_US |
Published: |
Washington, DC: World Bank
2005
|
Subjects: | FINANCIAL ANALYSIS; STATISTICS; MODELS; ERROR CORRECTION MODELS; COINTEGRATION; SWAP TRANSACTIONS; TREASURY BONDS; YIELD INCREASE; MORTGAGE-BACKED SECURITIES; ANALYTICAL METHODS;, |
Online Access: | http://documents.worldbank.org/curated/en/2005/06/6258950/determines-swap-spreads https://hdl.handle.net/10986/7272 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | This paper examines the evolution of the
U.S. interest swap market. The authors review the theory and
past empirical studies on U.S. swap spreads, and estimate an
error-correction model for maturities of 2, 5, and 10 years
from 1994 to 2004. Financial theory depicts swaps as
contracts indexed on London Inter-Bank Offered (LIBOR)
rates, rendered almost free of counterparty default risk by
mark-to-market and collateralization. Swap spreads reflect
the LIBOR credit quality (credit component) and a liquidity
convenience premium present in Treasury rates (liquidity
component). Multifactor models that were estimated on
observed swap rates highlighted the central role played by
the liquidity component in explaining swap-spread dynamics
over the past 15 years. The multifactor models also found
some puzzling empirical results. Statistical models, on the
other hand, based mainly on market analysis, faced technical
difficulties arising from the presence of regime changes,
from the non-stationary in swap spreads, and from the
coexistence of long-term and shorter-term determinants.
Against this background, the authors apply an
error-correction methodology based on the concept of
co-integration. They find that U.S. dollar swap spreads and
the supply of U.S. Treasury bonds are co-integrated,
suggesting that the Treasury supply is a key determinant on
a long-term horizon. The authors estimate an
error-correction model that integrates this long-term
relationship with the influence of four shorter-term
determinants: the AA spread, the repo rate, the difference
between on-the-run and off-the-run yields, and the duration
of mortgage-backed securities. The error-correction model
fits observed swap spreads quite well over the sample
period. The authors illustrate how the same model can be
used to carry out scenario analysis. |
---|