Ebook Mind the Gap: Disentangling Credit and Liquidity in Risk Spreads
The period since August 2007 has been marked by an unusual widening of risk spreads and increased volatility in financial asset prices. One example is the spread between term interbank borrowing rates and overnight indexed swap (OIS) rates of comparable maturity. These spreads have risen from 4 basis points to more than 150 basis points at their one and three month maturities, reaching their widest levels since the inception of the OIS market. Spreads between different euro-area government bond yields have also widened dramatically over this period. For instance, the spread of the Italian ten-year government bond yield over its German counterpart rose from 20 basis points before August 2007 to over 160 basis points, reaching its highest level since the inception of the euro.
The rise in spreads could represent heightened perceived default risk or greater compensation demanded by risk averse investors against the risk of default. Alternatively, it could represent a premium demanded by investors to induce them to hold comparatively illiquid assets. Decomposing interest rate spreads into default risk and liquidity components is important for investors’ portfolio allocation decisions investors with the longest investment horizons should prefer to hold higher yielding assets if these elevated yields represent compensation for poor liquidity, but not if they represent a greater risk of default. Amihud and Mendelson (1986) and Long staff (2005) both propose models with different types of investors, in which the longer-horizon investors hold less liquid assets and receive a premium for doing so. The default versus liquidity decomposition is also important for policymakers if rising spreads primarily represent the effects of poor liquidity, then policy actions to improve market functioning could help to dampen spread widening. For example, an exchange of more liquid bonds for their less liquid counterparts could promote more orderly market functioning. On the other hand, if the widening largely represents a heightened risk of default, then only actions aiming to improve the solvency of the banks/governments in question will ultimately work.
This paper makes three contributions in decomposing risk spreads into credit and liquidity components. My first contribution is to construct new measures of credit and liquidity. My credit measure is a time varying indicator of credit tiering, defined as the spread between the actual transactions rates paid by banks that are good credit risks and rates paid by those that are bad credit risks. I construct this variable with high frequency data from e-MID, an electronic interbank deposit trading platform which provides a unique identifier for the borrowing bank in each transaction. Each participant on the platform has identical search costs, so the fact that different banks pay different prices to borrow at the same point in time can only reflect counterparty credit risk.
My new market liquidity measure is the yield spread between German federal government bonds and KfW (Kreditanstalt für Wiederaufbau) agency bonds. I construct this variable with high-frequency data from MTS, an electronic trading platform for euro-area bonds. KfW bonds are German agency bonds that are less liquid than their federal government counterparts, but that are also fully guaranteed against default by the German federal government and so have identical cash flows. The spread between these two bonds is a pure measure of the German government bond market liquidity premium demanded by investors.
My second contribution is to use these credit and liquidity measures to explain the widening of interest rate spreads and to provide empirical evidence for general equilibrium relations between asset market liquidity and money markets. Over my sample, my credit and liquidity measures are nearly orthogonal with each other, enabling me to econometrically identify their separate effects on interest rate spreads. I find that both credit and liquidity effects are important in explaining the widening of LIBOR-OIS spreads, but that my measure of market liquidity explains a greater share of the widening. This result supports the idea that asset market liquidity directly affects money markets, providing empirical evidence for models that relate asset market liquidity to money markets such as Brunnermeier and Pedersen (2008) and Bolton, Santos, and Sheinkmann (2008). Turning to sovereign bond spreads, my measure of market liquidity explains 77 percent of the sovereign spread levels and 68 percent of their widening since August 2007. The large role of liquidity is perhaps surprising because previous empirical work on decomposing risk spreads has generally found a much smaller role for liquidity (see for example, Beber, Brandt and Kavajecz (2006); Longstaff, Mithal and Neis (2005); Taylor and Williams (2008); and McAndrews, Sarkar and Wang (2008), discussed later in this paper).
The difference between my findings and those of the existing literature is due to my improved measures of credit and liquidity. My liquidity measure is not contaminated by any effects from credit, and vice versa. Moreover, it reflects the cost of transacting today as well as compensation for the risk that transactions costs in less liquid markets may rise further in the future. The latter channel is missed by existing liquidity measures. Liquidity risk represents the possibility that liquidity will dry up in the future at exactly the time when investors’ marginal utility is at its highest. A bond could have low transactions costs today, but still have high liquidity risk. My third contribution—to investigate this liquidity channel further—is to construct a measure of liquidity risk defined as the covariance between a bond’s return and a measure of bond market liquidity, drawing on the theoretical models of Acharya and Pedersen (2005), and Pastor and Stambaugh (2003).
The plan for the remainder of this paper is as follows. In Section 2, I introduce the data sets that I use to construct my new credit and liquidity measures and the interest rate spreads that I will parse into these components. In Section 3, I detail my construction of new liquidity and credit measures, and describe how they are distinct from each other. In Sections 4 and 5, I discuss the determinants of the interbank and sovereign debt market spreads, respectively. In Section 6, I discuss the effect of liquidity risk on interest rate spreads. Section 7 concludes.
Download
PDF Ebook Mind the Gap: Disentangling Credit and Liquidity in Risk Spreads
Posted in :
