The Choice: Chapter 11 or Exchange Offer?

• Large firms with public debt in financial distress must restructure — but how?
  • Chapter 11: court-supervised reorganization under the U.S. Bankruptcy Code.
  • Exchange offer: out-of-court exchange of old debt for new securities (debt, equity, or cash).
• The choice has major consequences for all stakeholders:
  • Who recovers how much, and in what form.
  • Whether equity survives or is wiped out.
  • How long the process takes and at what cost.
  • Whether the firm continues as a going concern.
• What determines this choice?
  • Prior work focuses on financial characteristics of the firm.
  • We ask whether the equity and bond holdings of institutional investors are associated with the restructuring choice.

Why Ownership Should Matter

• Equity holders and bondholders have competing interests over restructuring.

• Equity holders prefer Exchange Offers:

  • Equity is a call option → favor continuation and risk-taking outside bankruptcy.
  • Absolute priority in Chapter 11 often wipes out equity (though APR violations are documented).
  • Exchange offer preserves some equity value and avoids wipeout.

• Bondholders face conflicting forces:

  • Secured creditors: Ch. 11 enforces priority → favors court.
  • Unsecured creditors: EO avoids bankruptcy costs and Ch. 11’s secured-creditor squeeze → favors out-of-court.
  • Both: Ch. 11 curbs equity risk-taking that erodes recovery → favors court.

Institutional Investors Dominate Both Sides

• Holdings:
  • Institutional investors hold 70–80% of publicly traded U.S. equities (Pensions & Investments 2017).
  • Institutional investors hold >90% of U.S. corporate bonds (Koijen and Yogo 2022).
  • Bonds account for 60–70% of outstanding debt at large U.S. firms (Bochner et al. 2020).
• But institutional investors are not monolithic:
  • Pension funds vs. hedge funds: different regulations, time horizons, investment mandates.
  • Insurance companies vs. private equity: different risk tolerance and portfolio objectives.
  • Banks vs. investment advisors: different regulatory constraints.
• Investor-type heterogeneity in equity and bond holdings may be a first-order determinant of restructuring outcomes

What We Know from Prior Work

• However, no study considers the joint effect of financial characteristics with contemporaneous bond and equity holdings across all institutional investor types
  • If bond holdings matter, not including them is an omitted variable problem
  • Fisher et al. (2026) extends this framework to a three-way multinomial choice (freefall vs. prenegotiated Ch. 11, with EO as base)

Our Contribution

• Novel dataset: equity and bond holdings for 15 investor types at 269 distressed firms (2000–2018)
  • Bond holdings hand-collected from Bloomberg/TRACE — no mandatory reporting for bonds
• Both equity and bond ownership strongly associated with restructuring choice
  • Substantial heterogeneity across investor types — not visible in single-asset studies
• VC/PE (probit) and hedge funds (elastic net) show opposite preferences across equity vs. bonds
  • Only detectable in a joint analysis
• Financial health, not debt composition, predicts the restructuring path
  • Prior debt structure findings may reflect omitted ownership variables

Chapter 11: Mechanics and the Secured-Creditor Shift

• The process:

  • Automatic stay + DIP financing keep operations running; creditor committees formed
  • Plan requires 2/3 approval by value and 1/2 by number per class; court confirmation; typically 1–2 years

• Four filing types — all included in our analysis: freefall, prenegotiated, prepackaged, §363 sale type definitions

• The shift toward secured creditors (Baird and Rasmussen 2002, 2003, 2006; Skeel 2003)

  • DIP financing gives banks leverage; Ch. 11 increasingly squeezes unsecured creditors and equity
  • For us: secured and unsecured bondholders face opposite incentives → bondholder-type composition predicts restructuring choice

Exchange Offers: Mechanics and the Holdout Problem

• The process:
  • Firm offers to exchange outstanding bonds for new securities (new debt at lower face value, or equity)
  • Each bondholder independently decides whether to tender
  • Trust Indenture Act: payment terms require unanimous consent; if a supermajority tenders, firm may strip remaining covenants via exit consent
• Impediments (Baird 1986; Roe 1987; Gertner and Scharfstein 1991; Hege and Mella-Barral 2019)
  • Holdout problem: bondholder refuses to tender hoping others exchange — worsens with more bondholders and heterogeneous expectations
  • Asymmetric information and complex debt structures multiply coordination failures

The Distressed Debt Market and Strategic Investors

• A $500B+ market has developed for distressed corporate debt (Harner 2008a, 2008b)

• 66% of distressed debt investors believe purchases are used to influence board or management decisions (Harner 2008b)

• Hedge funds acquire unsecured bonds pre-reorganization to:

  • Gain voting power in Ch. 11 plan approval
  • Exchange bonds for new equity in reorganized firm (Jiang et al. 2012)
  • Block exchange offers and force a court-supervised process

• Implication for our paper: investor holdings before restructuring are not random — strategic positioning may both predict and influence the restructuring choice

Sample Construction

• Step 1 — Identify distressed firms (Demiroglu and James 2015; variant of Gilson 1989; Gilson et al. 1990)
  • U.S. listed firms, 2000–2016; exclude utilities and financials
  • Bottom 5% of CRSP three-year cumulative stock return each calendar year
  • Exclude if: book leverage < 30%, interest coverage > 3, avg. assets < $100M (2002 dollars)
  • 840 distressed firm-years identified
• Step 2 — Identify restructuring events (2-year window around distress)
  • Chapter 11: LoPucki Bankruptcy Research Database, BankruptcyData.com, Moody’s MDRD, Fitch, S&P, EDGAR 8-K, Capital IQ
  • Exchange offers: Factiva (“exchange offer”, “debt restructuring”), EDGAR 8-K, Moody’s MDRD; (Danis 2016) and (Bratton and Levitin 2017) samples

Final Sample

• After exclusions:

• 269 firms total, 2000–2018
• 35 “dual procedures” (EO followed by Ch. 11 within 2 years); avg. gap = 277 days
• Four sectors account for ~75% of both samples: energy, communication services, consumer discretionary, industrials
• All but one Ch. 11 firm in distress at filing; 72% of EO firms

Filings by Year

Data Sources

The Investor Universe

• 8,214 unique investors: 1,370 bond holders, 6,844 equity holders
• 933 simultaneous bond+equity holders at Ch. 11 firms; 1,563 at EO firms
• 15 standardized investor types mapped from Bloomberg and Capital IQ nomenclature:
  • For equity: Bank, Corporation, Government, Hedge fund, Individual, Insurance company, Investment advisor, Pension fund, VC/PE + Rest (Brokerage, Endowment, Family trust, Foundation, Other, Sovereign wealth fund)
  • For bonds: Bank, Corporation, Hedge fund, Insurance company, Investment advisor, VC/PE + Rest (Brokerage, Endowment, Family trust, Foundation, Government, Individual, Other, Sovereign wealth fund)
• Bond investor universe (by count)
  • Insurance 47%, Investment advisors 34%, Hedge funds 11%
• Equity investor universe (by count)
  • Individuals 45%, Investment advisors 31%, Hedge funds 11%

Exchange Offer Firms Are Larger and Less Stressed

• EO firms roughly 80% larger by assets; much stronger coverage ratios
• Ch. 11 firms are deep in distress; EO firms are structurally distressed but operationally viable debt structure liability breakdown

Bond Information: Default Rate Tells the Story

• Default rate is the sharpest divide: 88% of Ch. 11 bonds in default vs. only 11% for EO firms
  • Bond exchange offers happen before crisis — a proactive restructuring, not a last resort
• Contrast with Demiroglu and James (2015): 63% default rate for out-of-court bank loan workouts
  • Bond EOs occur much earlier in distress than bank loan restructurings — qualitatively different phenomenon
• EO firms also have more bondholders (72 vs. 44), yet still coordinate successfully

Equity Holdings: Exchange Offer Firms Attract More Institutional Ownership

• Total equity holdings in EO firms significantly higher (70–74% vs. 54–65% across quarters)
  • Consistent with lower institutional presence at Ch. 11 firms
  • Two caveats: (i) reporting may weaken as firms near bankruptcy; (ii) when an investor exits, shares transfer to an unobserved buyer — we cannot confirm the buyer is non-institutional

Bond Holdings: EO Firms Have More Institutional Bond Holders

• Investment advisors, insurance companies, corporations hold significantly more bonds in EO firms
  • Reporting caveat: bond positions are not mandatory disclosures — Ch. 11 bondholders may have less incentive to report
• Hedge funds hold more bonds in Ch. 11 firms (though difference not significant) — consistent with strategic debt acquisition (Jiang et al. 2012)

Holdings Dynamics: Relative Changes

Probit Framework

• Outcome: Y_i = 1 if exchange offer, Y_i = 0 if Chapter 11

\Pr(Y_i = 1) = \Phi\bigl(\mathbf{x}_i' \boldsymbol{\beta}\bigr)

• Variable groups (all measured in the four quarters prior to filing):
  • Equity holdings by investor type: 10 types + dual holder measures + concentration (Top 5)
  • Bond holdings by investor type: 6 types + dual holder measures + concentration
  • Bond characteristics: number of bonds, bondholders, subordinate/secured bonds
  • Financial variables (at Q1): assets, ratios, profitability
  • Debt structure (at Q1): composition, seniority, security
  • Pre/During/Post 2008/09 crisis time dummies
• 39 variables total per specification; pseudo-R^2 = 0.41–0.43 across all three.

Variable Timing

• Q1, Q2, Q3, Q4 = 1, 2, 3, 4 quarters prior to filing (Q1 is the quarter immediately before)

• Financial and debt variables measured at Q1

  • Most recent snapshot before restructuring; consistent with the literature’s use of the most recent annual report

• Holdings measured at Q2, Q3, and Q4

  • Bond holdings drop sharply from Q2 to Q1 as firms approach default — Q1 bond data subject to selection bias
  • No theoretical prior on which quarter is most informative → run three separate specifications, one per quarter; results consistent across all three

Addressing Endogeneity

• Reverse causality concern: investors may change holdings in response to anticipated restructuring choice

• Why an IV approach is infeasible:

  • Need 16 instruments (10 equity + 6 bond types); each exclusion restriction must hold — almost certainly violated
  • Small sample (n = 269) → weak instrument bias severe

• Our approach: descriptive analysis, not causal inference

  • Document associations between ownership and restructuring choice
  • Ownership changes slowly → reverse causality less severe in practice
  • Results consistent across Q2, Q3, Q4 and across probit + elastic net

Equity Holdings → EO: Governments and Pension Funds

• Largest marginal effects in the model — a 1 pp increase in government equity ownership raises exchange offer probability by ~9 pp
• Pension funds and government pension plans have fiduciary duties to beneficiaries → must act in their best interest → avoid Ch. 11 wipeout of equity
• Government equity (state pension funds: California, Michigan, New York, Texas, Wisconsin) may reflect non-financial motives: preserving jobs, maintaining key industries → strong preference to avoid court bankruptcy
• Both types face regulatory scrutiny on portfolio losses → motivated to avoid the negative return associated with equity elimination in Ch. 11

Equity Holdings → EO: VC/PE, Hedge Funds, Individuals, Investment Advisors

• VC/PE firms: highly significant across all quarters — quick EO resolution → lower holding costs, faster capital redeployment; contrast with VC/PE bond holdings (strongly negative, see below)
• Hedge funds: positive at Q2–Q3 — equity stake → prefer out-of-court to preserve value; consistent with Lim (2015)
• Individuals (insiders/management): prefer to avoid court, preserve control
• Investment advisors: largest equity holder group; fiduciary duties favor value-preserving exchange offer

Equity Holdings: Banks and Insurance Companies (No Significant Effect)

• Neither is statistically significant in the probit — surprising given their regulatory constraints RWA / RBC
• Banks: distressed equity is capital-intensive under RWA rules → would expect EO preference, but no evidence
  • Likely explanation: stakes are small, inherited from prior loan-to-equity conversions — not strategically placed
• Insurance companies: same NAIC RBC logic → same EO expectation; elastic net finds negative sign (6/9) → tilts toward Ch. 11
  • Similarly: small, passive stakes not driven by restructuring preferences

Bond Holdings → EO: Corporations and Investment Advisors

• Corporate bondholders — largest marginal effect among bond variables
  • May have strategic business relationships (suppliers, customers, JV partners) → prefer out-of-court to preserve counterparty
  • Court-supervised Ch. 11 creates uncertainty around contracts and ongoing business relationships
• Investment advisors: positive — consistent with equity holding preference; fiduciary orientation → minimize recovery risk → prefer EO

Bond Holdings → Ch. 11: VC/PE Firms (and Hedge Funds via Elastic Net)

• VC/PE: strong, consistent, significantly negative — loan-to-own strategy: buy bonds at a discount, use Ch. 11 to exchange for new equity (Jiang et al. 2012); absolute priority enforced by court beats negotiated EO terms
  • The VC/PE paradox: opposite preferences across equity vs. bonds — only visible in joint analysis
• Hedge funds: not significant in probit; elastic net finds negative sign in all 9 regressions → preference for Ch. 11 as bond holders confirmed
  • Hedge funds as equity holders: preserve value → EO; as bond holders: gain voting power → Ch. 11 (Jiang et al. 2012)

Dual Holdings: A Puzzle

• Both measures negatively associated with exchange offer (conditional on investor type composition)
  • Univariate: dual holders more prevalent at EO firms univariate
  • Multivariate: controlling for investor-type composition, dual holders prefer Ch. 11
• Bond ownership effect dominates at Q2 (% bonds held by equity investors)
• Equity ownership effect dominates at Q3–Q4 (% shares held by bond investors)
• Conditional on investor type: simultaneous holders may prefer Ch. 11’s structured priority rules → enforcing claims via court more reliable than voluntary exchange
  • Contrast with Chu et al. (2025): they do not control for investor-type composition

Capital Structure Complexity and Ownership Concentration

• More bonds (negative): greater coordination problem → holdout risk increases → Ch. 11 more likely
  • Each additional bond class requires separate negotiation in an exchange offer
• More subordinate bonds (positive): unsecured bondholders most exposed to losses in Ch. 11 → strongly prefer EO to avoid elimination under absolute priority
• Top 5 equity concentration (negative): may reflect large activist/block holders positioned to benefit from Ch. 11 process (DIP lending, credit bidding)

Financial Controls: Robust Across All Specifications

• Consistent across all three quarters — independent of how ownership is measured
  • Short-term solvency pressure (current LTD) → Ch. 11 (immediate creditor protection needed)
  • Balance sheet strength (assets/liabilities) → EO (firm has value to preserve)
  • Negative EBIT margin → Ch. 11 (operating losses compound financial stress)
  • Higher interest coverage → EO (more viable as going concern)
• Debt structure variables (bonds/notes share, senior, bank, unsecured): not significant — composition doesn’t predict choice, only the aggregate financial health does

Elastic Net Variable Selection

• Why elastic net?
  • 39 variables, n = 269 → overfitting risk; potential multicollinearity
  • Elastic net combines LASSO (L_1) and ridge (L_2) penalties: enforces sparsity while handling correlated predictors (Zou and Hastie 2005) details
  • Algorithm-driven selection: no subjective variable elimination
• Setup:
  • 9 combinations of equity holdings quarter × bond holdings quarter (Q2, Q3, Q4 each)
  • Each run starts from all 39 variables in the probit; elastic net selects a subset
  • Report: which variables are selected and the sign of their coefficient
  • 26 of 39 variables selected at least once; 19 selected on average per regression

Key Elastic Net Results

Sign Consistency: Elastic Net Endorses Probit

• For probit-significant variables: signs never reverse in elastic net
• Elastic net selects ownership variables robustly — not artifacts of overfitting
• Financial variables selected in all 9 regressions — significant after controlling for ownership
• Elastic net clarifies probit non-results:
  • Hedge fund bonds: selected 9/9 with negative sign → hedge funds prefer Ch. 11 as bond holders
  • Insurance company equity: selected 6/9 with negative sign → clarifies the non-result in probit

Findings

• Equity ownership → Exchange Offer for most types:

  • Governments and pension funds: largest effects (~7–10 pp per 1 pp increase)
  • VC/PE firms, hedge funds, individuals, investment advisors: positive
  • Banks and insurance companies: no significant association

• Bond ownership → heterogeneous across investor types:

  • Corporations and investment advisors → Exchange Offer
  • VC/PE firms → Chapter 11 (loan-to-own strategy)
  • Hedge funds → Chapter 11 (elastic net only; Jiang et al. (2012) loan-to-own via bonds)

• Key cross-asset finding: VC/PE (probit) and hedge funds (elastic net) show opposite preferences across equity vs. bonds — only detectable in joint analysis

• Elastic net confirms all signs across all 9 quarter combinations — results are not artifacts of overfitting

Contributions & Implications

• Novel joint equity + bond holdings dataset
  • First to study restructuring with all investor types simultaneously; future work must account for joint holdings to avoid omitted variable bias
• Cross-asset heterogeneity is first-order
  • Single-asset studies are confounded; joint analysis is necessary to detect investor-type preferences
• Financial health, not debt composition, predicts restructuring choice
  • Debt structure findings in prior work may reflect omitted ownership variables
• Capital structure complexity and the Ch. 11 system
  • More bond classes impedes EO (coordination frictions); VC/PE and hedge fund bond strategies consistent with the secular shift toward secured creditors in Ch. 11

Risk-Weighted Assets (RWA) and Restructuring Incentives

• Banks (Basel III RWA): capital requirement = risk weight × asset value
  • Distressed equity carries 100–150% risk weight → capital-intensive to hold
  • Lower losses under EO → smaller write-down → less capital consumed → expected EO preference
• Insurance companies (NAIC Risk-Based Capital (RBC)):
  • Risk-based capital requirements penalize low-rated or distressed securities
  • Same logic: EO preserves value → lower RBC charge → expected EO preference
• We find no evidence for banks; insurance companies tilt toward Ch. 11 in elastic net
  • Stakes are likely small and not strategically placed → RWA logic may not apply in practice

Short-term Debt Pressure vs. Long-term Debt Burden

• Ch. 11 firms face imminent payment obligations
  • Chapter 11’s automatic stay provides relief
• EO firms have a long-term debt burden they can restructure via exchange
  • Pattern consistent with (Chatterjee et al. 1996) and (Gilson et al. 1990)
• Firms often reclassify debt as current as default approaches
  • Inflates Ch. 11 current LTD / liabilities ratios

Debt Structure: Surprisingly Similar

• Debt composition is largely not predictive of restructuring choice — ownership and financial health are
• Univariate: 5 of 6 debt composition variables show no significant difference across groups
• Multivariate (probit): debt structure variables not significant in any specification

Bloomberg Bond Holdings: Data Collection

Holdings Dynamics: Absolute Changes (pp)

Dual Holdings: Simultaneous Equity and Bond Positions

• By both measures, simultaneous holders are more prevalent at EO firms (univariate)
• Consistent with Chu et al. (2025): dual loan-equity holders mitigate shareholder-creditor conflict → out-of-court
• Caveat: controlling for investor-type composition in the probit, dual holdings become negatively associated with exchange offers — we return to this

Chapter 11 Filing Types

• Freefall
  • No prior agreement; plan negotiated entirely in court; longest and most costly
• Prenegotiated
  • Informal pre-filing agreement with key creditors; Ch. 11 binds holdouts; reduces time and cost
• Prepackaged
  • Plan and votes completed before filing; court only confirms; fastest path (weeks)
• §363 sale
  • Assets sold via court auction free of liabilities; equity typically wiped out

Data Cleaning Rules

• Investor names — we merge same investor under different names, and subsidiary positions with parent (when linkable)

• Bond holdings: Quarterly gaps filled by carry-forward.

  • Semi-annual reporters (alternating zeros) and runs of 1–3 missing quarters filled from the prior non-zero value; no report at all → 0
  • In total, 275 Ch. 11 and 366 EO individual bonds corrected; 144A/REGS duplicates aggregated for Ch. 11 (no effect on results — holdings scaled to total reported value)

• Equity holdings: missing shares outstanding estimated from % of CSO or carried forward

  • 19 Ch. 11 and 63 EO firms had splits or new issuances corrected; reporting outliers and typos removed

• Financials — Compustat download cross-checked against Capital IQ and firm quarterly reports for all sample firms

Elastic Net: Estimation Details

• Penalized log-likelihood for binary outcome Y_i \in \{0,1\}:

\min_{\boldsymbol{\beta}} \left[ -\frac{1}{n}\sum_{i=1}^{n} \ell(Y_i, \mathbf{x}_i'\boldsymbol{\beta}) + \lambda \left( \frac{1-\alpha}{2} \|\boldsymbol{\beta}\|_2^2 + \alpha \|\boldsymbol{\beta}\|_1 \right) \right]

• \alpha = 0: ridge (shrink all); \alpha = 1: LASSO (exact zeros); we use \alpha = 0.5 for sparsity + stability under correlated predictors

• \lambda > 0: overall penalty strength — larger \lambda forces more coefficients to zero

• Selecting \lambda_{\min} via 10-fold cross-validation:

  • Partition sample into 10 folds; for each \lambda, train on 9 folds and measure out-of-sample prediction error on the held-out fold
  • \lambda_{\min}: the value that minimizes average out-of-sample prediction error across the 10 folds

• All continuous predictors standardized so \lambda penalizes all coefficients on the same scale