Yong Chen, Texas A&M University

Zheng Sun, University of California Irvine

Haibei Zhao, Lehigh University

​

CICF 2026: Information, Liquidity, and Asset Management

Reigniting the fire (sales): Evidence from corporate bond mutual funds

Motivation

• Fire sale: forced sale of an asset at a price that is dislocated from fundamental values (Shleifer and Vishny, 1992)
• Non-fundamental price pressure (NFPP) has important implications for
• Demand-based asset pricing
• Fragility and fund runs
• Feedback effects on corporate policies

Empirical evidence

• Mixed evidence of NFPP
• Yes, using fund flows as an instrument (e.g., Coval and Stafford, 2007; Greenwood and Thesmar, 2011; Ellul, Jotikasthira, and Lundblad, 2011; Jotikasthira, Lundblad, and Ramadorai, 2012)
• No, after controlling for fundamentals such as stock returns, bond ratings, issuer by time FEs (E.g., Ambrose, Cai, and Helwege, 2012; Wardlaw, 2020; Choi et al., 2020; Chaudhary, Fu, and Li, 2023)

Choi et al. (2020)

• Bonds issued by the same company are subject to the same fundamentals, but may differ in flow-induced selling pressure
• Matched controls: the same issuer, credit rating, seniority, and option features; less than 1 year of maturity difference; and least difference in age
• Issuer by time FEs
• Redemptions from CBMFs and the resulting sell-offs do not lead to asset fire sales, even in extreme market conditions

Our idea

• Standard event studies: best to use counterfactuals with near-identical characteristics
• Powerful controls to guard against false discoveries
• Fire sale tests: identical assets should be priced equally
• Price discrepancy will be arbitraged away so a non-effect here does not preclude NFPP: selling pressure may spill over from the treatment to the control, depressing prices of both!

The dilemma

Close substitutes as controls

• Treatment-control bonds of the same issuer
• Introduce arbitrage frictions
• Yield curve convergence risk (Greenwood and Vayanos, 2014; Vayanos and Vila, 2021)
• Optionality / Liquidity
• Speed (daily vs. monthly)
• Crisis periods
• Passive holders
• Result: NFPP of 1.66% using monthly data and 3.74% using daily data for a representative bond

Fundamentals

• What about fundamental differences?
• Retain issuer×time FEs to control for issuer-level fundamentals
• Use yield spreads to control for duration risk
• Maturity and rating
• Additional robustness rules out information story (return reversals, mutual fund trades, imputed arbitrage trades, search costs)

Data

• Mutual fund returns and characteristics: CRSP
• Bond fund holdings: Morningstar
• Bond characteristics: Mergent FISD
• Bond yields, returns and prices: TRACE
• Sample period: 2002Q3 to 2014Q4 for comparisons with Choi et al., 2020 (CHST)
• Extend to 2019Q4 for robustness

Replication of CHST

Add less substitutable bonds

Daily data

• Reasons to use daily data
• Bessembinder et al. (2009): daily bond returns provide much greater statistical power
• Daily data better capture the early stage of fire sale spillover
• All conditional on treatment and controls being traded over 2 consecutive days (no stale pricing)

Daily data

NFPP of 3.74% for a representative bond with 6-year duration

Localization of demand equilibrium (Vayanos and Vila, 2021): Price pressure diffuses more to bonds of similar maturity

Substitution swap trades

• Active arbitrage: buy (relatively) underpriced and sell overpriced peer
• Substitution arbitrage: existing owners of the overpriced assets switch to holding the underpriced ones (substitution swap)
• Classify insurance companies into active and passive (O'Hara, Wang, and Zhou, 2018)
• If the control bonds are more owned by active (passive) insurers, substitution swap trades are more (less) likely

Information story

• “Within-issuer” information story: fire-sold bonds receive information-base selling first, then other bonds
• Addition evidence suggests otherwise:
• Return reversals
• Imputed arbitrage trades
• Mutual funds trades
• Search frictions

Return reversals

• After experiencing poor returns, immediate return reversal observed on fire sold bonds

Return reversal – alternative story

• Perhaps, the company had poor fundamental one day then good fundamental next?
• No – after those days, the other bonds of the same company had WORSE returns
• What if bad fundamentals spill over to other bonds – inconsistent with the fire sold bond itself gets BETTER returns
• Close peer bonds on the yield curve receive a larger drag on price and distant peer a lower drag – perfectly consistent with limits to arbitrage due to yield convergence risk, not the information story

​

Imputed arbitrage trades

• Trader identity not observable; rely on imputed arbitrage trades
• Matched pairs of trading such that: (i) both trades are customer–dealer; (ii) the bond is purchased, and the other bond, belonging to the same issuer, is sold; (iii) the trades involve identical principal amount and occur within the same day; and (iv) neither trade is part of an imputed roundtrip trade
• Fire sold bonds have more IAT frequency (9.6%) and intensity (11.9%)
• Information story: sell on both fire sold bonds and other bonds, not buy

​

​

Imputed arbitrage trades

• More imputed arbitrage trades on fire sold bonds
• Particularly between fire sold bonds and their close peers (same company bonds with similar maturity)
• Consistent with term structure theories

​

Mutual fund trades

• Discretionary funds: flows rank between 30^th to 70^th (or funds with inflows)
• Trades by these funds are more information based
• Those funds buy both the fire sold bond and its peer bonds
• They do not view the fire sold bonds as fundamentally impaired
• They do not view the peer bonds of the affected issuer as fundamentally impaired

Search cost

• Feldhütter (2012): Price difference between large and small trades within the same bond identifies NFPP
• Large buyers get even better pricing during non-fundamental selling pressure
• We find large buyers receive an additional 20% discount when purchasing fire-sold bonds during the event quarter
• Fire-sold bonds experience strong reversals after the days when large buyers receive greater discounts
• Large buyers earn a premium for liquidity provision, rather than being adversely selected by informed seller

Liquidity management

• Fund cash holding is small after excluding Treasury bonds (12.4% down to 3.4%)
• Insufficient to weather heavy redemptions (flow at 25^th: −4%)
• Cash holdings helps reduce, but not eliminate, the necessity to sell bonds after outflows

Concluding remarks

• Find evidence of NFPP using close substitutes as controls
• Consistent with limits to arbitrage: bond substitutability, yield convergence risk, crisis period, daily data, insurance companies, return reversals, imputed arbitrage trades
• Ultimately, NFPP is a joint test of the fire sale hypothesis and the test assets
• “Strong form”: any non-fire-sold asset
• “Weak form”: non-fire-sold assets with (near) identical fundamentals
• “Semi-strong form”: non-fire-sold asset with similar fundamentals

​