6.8.6 Advanced Extensions

Basic Prerequisites: 02 · Illiquidity Measures and 04 · Liquidity Risk & Crises.

---

1. Intuition & Practical Objective

Equities have volume and continuous quotes; corporate bonds barely do. Bonds trade over-the-counter, infrequently, in opaque dealer markets, so the equity toolkit (ILLIQ, spreads) partially breaks - and yet bond illiquidity is larger and more crisis-sensitive than equities’. This page extends the asset-pricing-of-liquidity framework to the dealer/OTC world, where the frontier of the research now sits.

The extensions, in increasing sophistication:

1. Bao, Pan & Wang’s (2011) bond illiquidity $\gamma$ - a Roll-style measure built from transitory price noise (negative return autocovariance), usable with sparse price data. It shows bond illiquidity is huge (far larger than bid-ask spreads imply), strongly common, and spikes in crises.
2. Illiquidity in yield spreads - the market-level illiquidity component of corporate-bond yield spreads over Treasuries; in the 2008 crisis it overshadows the credit-risk component.
3. Flight to quality - when illiquidity spikes, investors flee bonds to Treasuries, widening the liquid-vs-illiquid spread; the very flight makes the illiquid side costlier.
4. Bridge to L-VaR - bond illiquidity feeds directly into liquidation-cost-adjusted risk (Liquidation Cost & L-VaR).

---

2. Mathematical Ground Truth & Derivations

The Bao $\gamma$ measure. Decompose a bond’s log price into a fundamental component and transitory noise ${\eta }_t$ (dealer inventory, search frictions, stale quotes):

$$\ln P_t=\ln F_t+{\eta }_t,{\eta }_t=\rho {\eta }_{t-1}+{\epsilon }_t,|\rho |<1.$$

The transitory component creates negative autocovariance in consecutive log-price changes, whose magnitude measures the transitory price noise = illiquidity:

$$\gamma =-\mathrm{cov}(\Delta \ln P_{t+1},\Delta \ln P_t)>0.$$

This is the same identification as Roll, but applied to the noise component rather than the bid-ask bounce. Bao, Pan & Wang (2011) find bond $\gamma$ is economically large - substantially bigger than quoted bid-ask spreads - confirming that bonds’ illiquidity is not primarily a spread story but a depth/search/market-making story. It also shows strong commonality across bonds and a sharp rise during 2007–08 (aggregate $\gamma$ roughly doubled by Aug 2007 and tripled by Mar 2008).

Illiquidity vs. credit in yield spreads. Decompose a bond’s yield spread over the risk-free rate into a credit-risk part and an illiquidity part. Bao et al. show that the aggregate illiquidity component explains a substantial share of the time variation of yield spreads, and that in the crisis it overshadows the credit-risk component - especially for high-rated bonds, where the sudden illiquidity jump was the story.

The flight-to-quality channel. Because illiquidity is common and spikes when risk premia rise, the price of the liquid benchmark (Treasuries) rises exactly when everything else becomes illiquid. The wedge between “safe-and-liquid” and “risky-and-illiquid” widens, and the covariance that determines the crisis liquidity premium (page 04) is precisely what widens it.

---

3. Computational Implementation - estimating bond $\gamma$

Simulate an OTC bond whose log price is a fundamental random walk plus AR(1) transitory noise, estimate $\gamma$ from the negative autocovariance of log-price changes, and show how it jumps when the transitory noise intensifies (a crisis):

Intensifying the transitory price noise by 5× raises measured bond illiquidity by ~16× - the same non-linear sensitivity Bao, Pan & Wang document around 2008: small increases in dealer stress produce dramatic jumps in measured illiquidity, and hence in the illiquidity component of yield spreads. This is the mechanism behind “spreads blow out in a crisis.”

---

4. Failure Modes & First-Principles Breakdowns

• Sparse data. Bonds trade irregularly; the autocovariance of observed price changes mixes the transitory noise with the irregular sampling gap. $\gamma$ is best computed on periods with reliable prices, and is sensitive to which trades you observe.
• Confounding with credit risk. Transitory price noise can reflect news-driven repricing as well as illiquidity. Bao et al. control carefully; a naive $\gamma$ overstates illiquidity for news-heavy names.
• Level vs. crisis state. Like every liquidity measure, $\gamma$ in calm times understates the crisis-state value (16× here); pricing bond liquidity risk requires tail-state estimates, not averages.
• The funding link. Bond illiquidity is inseparable from dealer balance sheets: when dealers cut inventory (post-GFC, in crises), $\gamma$ jumps. This ties directly into the funding-liquidity story.

---

5. References

• Bao, Pan & Wang (2011). The illiquidity of corporate bonds. Journal of Finance 66(3), 911–946. The $\gamma$ measure, commonality, the 2008 crisis spike, and its dominance over credit risk in spreads.
• Amihud, Mendelson & Pedersen (2013). Market Liquidity
• Acharya & Pedersen (2005). Asset pricing with liquidity risk. JFE 77
• Dick-Nielsen, Feldhütter & Lando (2012). Corporate bond liquidity before and after the onset of the subprime crisis. JFE 103
• O’Hara (2015). High frequency market microstructure. JFE 116

---

6. Connected Graph Bridges

• Back: 05 · Failure Modes & Practice · 04 · Liquidity Risk & Crises · Index Hub
• Sibling: Inventory Management & Quote Skewing (dealer market-making) · Liquidation Cost & L-VaR
• Forward: Liquidity Risk & Funding (the funding side of the same illiquidity)