Pith. sign in

REVIEW 3 major objections 5 minor 26 references

Fractional Claims Trades and Donations in Financial Networks

T0 review · 3 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that fractional claims trading lets one bank rescue a distressed creditor optimally, in polynomial time, even when defaults are costly.

desk verdict Fixable LP gap in the main positive theorem; otherwise a solid advance in fractional claims trading. read the letter →

arxiv 2502.06515 v1 pith:ZJEQ5VZ3 submitted 2025-02-10 cs.GT

Martin Hoefer , Lars Huth , Lisa Wilhelmi This is my paper · ORCID
classification cs.GT MSC 68Q2591B3091G80
keywords financialnetworksclaimstradingfractionaltradesdonationsdefaultcostsclearingstatesParetoimprovementNP-hardness
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper asks whether one bank can rescue a distressed creditor by buying parts of its incoming claims—an operation recognized in the U.S. Bankruptcy Code—and whether the best rescue can be found efficiently. It proves that when the buyer and seller may trade any fraction of a claim, an optimal rescue trade that strictly helps the creditor without hurting the buyer can be computed in polynomial time, even when defaults carry costs. The computed trade weakly improves every bank's assets relative to any other creditor-positive trade. The same guarantee extends to trading several incoming claims of one creditor. By contrast, when a single debtor's claims are sold to help its creditors, or when one bank donates to several banks, maximizing the improvement is NP-hard once default costs are present, though donations without default costs remain efficiently solvable.

What carries the argument

The load-bearing object is the default hierarchy of a split network $F^s(v,w)$, in which $v$ and $w$ are each split into a sink that receives incoming edges and a source that pays outgoing edges, with external assets routed to the sources. As the post-trade assets $a'_v = a^{\mathrm{out}}_v$ vary, the set of solvent banks changes only at finitely many breakpoints; the hierarchy lists at most $n$ intervals $[a^{(j+1)}_v, a^{(j)}_v)$ with their invariant solvent sets $S^{(j)}$. Proposition 5 computes the hierarchy in polynomial time via a modified clearing-state routine and LP (1), whose optimal value gives the next breakpoint even though its solution is not necessarily the clearing state. For the trade itself, the regime constraints plus the trade variables $\alpha, \beta$ are nonlinear, but Proposition 2's normalization (either $\beta = 1$ or $\rho = a^x_w$) and the substitution $y = r'_u \beta \ell_e$ turn them into linear programs, one per hierarchy level. For multi-trades of incoming edges, Lemma 10 shows an optimal trade needs at most one fractional claim, chosen in sorted order of debtor recovery rates; within each hierarchy level payments are linear in $a'_v$, so only $O(k^2)$ orderings need enumeration.

What would settle it

Run the default-hierarchy computation from Proposition 5 on a three-bank network with default cost $\delta < 1$ and compare its breakpoints against the actual clearing states obtained by enumerating all fixed points for a dense grid of external assets for $v$; any interval $[a_v^{(j+1)}, a_v^{(j)})$ in which the insolvent set differs from $S^{(j)}$ would disprove Proposition 5 and the algorithm's coverage guarantee.

Watch

Extended reading notes

Core claim

The paper's central claim is that fractional claims trading is efficiently optimizable exactly where a rescue is possible. Fixing an Eisenberg-Noe financial network with arbitrary default cost, a creditor $v$, a buyer $w$, and one incoming claim $(u,v)$—or any number of incoming claims of $v$—if some trade makes $v$ strictly better off and $w$ no worse off, then a polynomial-time algorithm finds a trade whose post-trade assets are at least as high as those of every such creditor-positive trade, for $v$, $w$, and every other bank. In networks without default cost, this specializes to the optimal creditor-positive trade; with default cost, the computed trade is a weak Pareto improvement over that benchmark and may even be a positive trade in which both $v$ and $w$ improve. Fractionality is essential: the paper's own example shows a three-quarter trade beats the best binary trade. The same framework yields NP-hardness when the operation instead redistributes a common debtor's outgoing claims to multiple creditors, or one bank donates to multiple banks, in networks with default costs; without default costs, optimal multi-donations are polynomial-time solvable.

Load-bearing premise

The whole construction depends on the standard assumption that the clearing state is the maximal fixed point of the payment equations and that it moves monotonically downward when a bank's external assets shrink; if either fails, the default hierarchy's intervals no longer cover all possible post-trade outcomes.

Editorial extensions

If this is right

  • A court or regulator can compute the best fractional rescue of a specified creditor in polynomial time, or certify that no rescue trade exists, instead of searching combinatorially.
  • The trade produced by the algorithm is a weak Pareto improvement: every bank in the network receives at least as many post-trade assets as under any creditor-positive trade.
  • In networks without default costs, the optimal creditor-positive trade—single or multi-claim—is polynomial-time solvable, and so are optimal multi-donations; this contrasts with the earlier binary-trade setting where multi-trades are NP-hard.
  • With default costs, maximizing asset gains to a common debtor's creditors or from multi-donations is NP-hard, so exact efficient optimization for those operations is not expected unless P equals NP.
  • The same machinery handles buyer-improved trades targeting a level $\omega > a_w$ for the buyer's assets, as stated in Corollary 14.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial extension: the default-hierarchy and LP approach should transfer to other single-bank asset shifts such as debt forgiveness or prepayments, predicting polynomial-time Pareto-optimal interventions in those settings as well—a hypothesis that can be tested by adapting LPs (1) and (2).
  • Editorial extension: the tractability boundary appears to be drawn by whose claims move, not by fractions: moving a single creditor's incoming claims is easy, while reallocating one debtor's outgoing claims across multiple creditors is hard, suggesting other multi-creditor interventions will also be NP-hard under default costs.
  • Editorial extension: because Lemma 10 reduces an optimal multi-trade to one fractional claim plus full transfers of cheaper claims, a practical implementation could quote each rescue as a short menu rather than a continuous set of contracts.
  • Editorial extension: the monotonicity premise could be stress-tested empirically on real interbank liability data by comparing default-hierarchy predictions against simulated clearing under many small external-asset shocks; failures would mark a domain boundary for the algorithm.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper studies fractional claims trades and donations in the Eisenberg–Noe financial network model, with and without default costs. For a distressed creditor v and a buyer w, it introduces creditor-positive trades (v strictly improves, w is not harmed) and shows that, when a single claim or a multi-set of claims of v is traded, a trade weakly Pareto-dominating every creditor-positive trade can be computed in polynomial time. The main technical tool is a default hierarchy that partitions possible post-trade assets of v into intervals with invariant insolvent sets. For trades of outgoing edges of a common debtor, the paper proves NP-hardness in networks with default costs and gives polynomial-time algorithms for excess-return trades and donations in networks without default costs. The central positive results are Theorems 6 and 8, with Corollaries 7 and 9 for the default-cost-free case.

Significance. If the main results are correct, the paper establishes a clean algorithmic frontier: trading a creditor's incoming claims is efficiently optimizable even with default costs, while trading a debtor's outgoing claims is NP-hard under default costs unless restricted to excess returns. The paper ships explicit LP formulations and reductions, and the polynomial-time claims are falsifiable and concrete. However, the proof of Theorem 6 contains a missing feasibility constraint in the main LP substitution; since Theorem 8 inherits this step, the central algorithmic claim is not yet established. The conceptual framework and the hardness reductions are otherwise coherent, and the paper is a substantive contribution to the algorithmic financial-networks literature if the gap is fixed.

major comments (3)
  1. [§3.2, proof of Theorem 6, Eq. (5)] In the case ρ = ax_w and r'_u > 0, the substitution y = r'_u β ℓ_e discards the model constraint ρ ≤ β ℓ_e. With ρ = ax_w, this constraint reads ax_w ≤ β ℓ_e, i.e. y ≥ r'_u · ax_w. The displayed LP (5) only contains y ∈ [0, r'_u ℓ_e], so it admits solutions with y < r'_u · ax_w that correspond to β < ax_w / ℓ_e and hence to an invalid trade whose return exceeds the traded liability. Because the objective maximizes a'_v and a'_v decreases in y, an LP optimum can prefer exactly such an infeasible donation-like solution. The fix is to add the linear constraint y ≥ r'_u · ax_w to (5). This is load-bearing: Theorem 8 invokes Theorem 6 as a subroutine.
  2. [§3.2, proof of Theorem 6, subcase r'_u = 0] In the subcase r'_u = 0, the proof asserts that the constraints become linear, but it still must enforce the feasibility condition ρ ≤ β ℓ_e. Since ρ = ax_w and r'_u = 0, the traded liability is β ℓ_e, so the required constraint is ax_w ≤ ℓ_e (using β ≤ 1). Equation (5) as written does not contain this condition, so the LP can output a trade that pays more than the liability of the traded claim. The proof should explicitly add ax_w ≤ ℓ_e when r'_u = 0, or otherwise derive a suitable linear constraint.
  3. [§3.2, proof of Theorem 6, after Eq. (5)] The statement 'α = r'_u ax_w' is dimensionally inconsistent: α is a dimensionless haircut rate while r'_u ax_w has units of money. The correct relation is α = r'_u ax_w / y, and the factor 1/y disappears only because y is defined as r'_u β ℓ_e. Since α is eliminated in (5), this is a presentation issue, but the underlying missing lower bound on y identified above is not merely cosmetic; it is the substantive gap in the proof.
minor comments (5)
  1. [§3.1, Definition 4] The hierarchy writes a(0)_v = ∞ > L_v = a(1)_v, but the subsequent intervals and LP computations treat a(0)_v as a real breakpoint; it would be clearer to say that a(0)_v is a formal upper bound and that all relevant values satisfy a(0)_v = L_v for the purpose of interval enumeration.
  2. [§3.1, proof of Proposition 5] The proof uses the notation S_i, S_{i+1}, and S_i \ S_f in the paragraph after Algorithm 1 without defining these sets; they presumably refer to S(j) and the frontier set S_f. Please align the notation with Definition 4.
  3. [§3.2, proof of Theorem 6, Eq. (4)] In the definition of δ_v, the text says 'δ_v = 1 for i = 0' but no index i has been introduced; this should be a condition on the interval j or on a'_v ≥ L_v.
  4. [§5, Figure 2] The caption and proof refer to edge liabilities M−|S_i| and M−|S_k|; for clarity, state explicitly that |S_i| denotes the cardinality of set S_i and that all other displayed unlabeled edges have unit liability.
  5. [§6, Corollary 14] The phrase 'buyer-ω-improved multi-trade' is used before it is formally defined in the same paragraph; a one-sentence definition of the class of benchmark trades would improve readability.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: Theorems 6 and 8 derive fractional-trade guarantees from first-principles LP enumeration over the default hierarchy against external benchmarks [5,21]; same-author citations appear only in supporting roles.

full rationale

Circularity walk: the paper's central results (Theorems 6 and 8) derive a domination guarantee over the optimal creditor-positive trade from exhaustive enumeration, not from a definitional identification. The default hierarchy (Def. 4, Prop. 5) partitions possible post-trade assets of v by the solvency sets of the split network Fs(v,w); its correctness rests on the supremal-fixed-point convention and monotonicity of the Eisenberg-Noe clearing state, imported from external works [5,21]. In Theorem 6, for each hierarchy interval an LP maximizes a'_v subject to fixed-point and trade-feasibility constraints; the domination argument is that the optimal creditor-positive trade itself is feasible for the LP of its interval, so the LP optimum is an upper bound. The objective is not defined in terms of the benchmark, and no fitted parameter is renamed as a prediction; there is no self-definitional or fitted-input pattern. Self-citations are present but supporting: Prop. 1 extends the binary no-positive-trade result of [7] (Hoefer-Ventre-Wilhelmi) to fractional trades via a self-contained multi-edge splitting reduction; the Pareto-improvement assertion cites [6,7,18] with the external [18] (Papp-Wattenhofer); the linearity of recovery rates in a hierarchy part (Thm. 8) cites [6,18], externally corroborated and independently verifiable as a linear-system fact. None of the main algorithmic claims reduces by construction to these citations. Donations-as-trades is an explicit modeling equivalence, not a renaming of a known result; Section 6's caveat that the computed trade dominates the benchmark class rather than globally optimizing is an honest limitation, not a circularity. Flagged but non-circular: Section 3.2's substitution (5) omits the model requirement ρ≤βℓ_e (equivalently y≥r'_u·ax_w when ρ=ax_w and r'_u>0), so an LP optimum may be an invalid trade; this is a soundness gap in the proof, not a circular reduction, and does not raise the circularity score. Verdict: no significant circularity; score 1 reflects only minor same-author citations in supporting roles.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claims rest on the standard Eisenberg-Noe fixed-point model, monotonicity of the maximal clearing state, prior binary-trade impossibility, and the default-cost extension; no free parameters are fitted and no new entities are postulated. The main proof gap is a missing LP feasibility constraint, not an assumption.

assumptions (4)
  • standard math The Eisenberg-Noe clearing state is the greatest fixed point of the payment equations and is polynomial-time computable.
    Used throughout as the outcome of a financial network; cited to [5,21].
  • domain assumption Payments are proportional to liabilities, and default costs reduce the assets of insolvent banks by a factor delta.
    This is the model definition in Section 2.1; all algorithms and hardness results depend on it.
  • domain assumption Return payments are bounded by the buyer's external assets and by the traded liability.
    Central to the LP constraints and to the identified proof gap in Theorem 6; stated in Section 2.1.
  • standard math In networks without default cost, no binary multi-trade of incoming edges can strictly improve both creditor and buyer.
    Used in Proposition 1 to rule out positive fractional trades; cited from [7].

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fractional Claims Trades and Donations in Financial Networks." pith.science (2026). https://pith.science/paper/ZJEQ5VZ3

@misc{pith2026250206515,
  author       = {Pith},
  title        = {Pith review of: Fractional Claims Trades and Donations in Financial Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZJEQ5VZ3}},
  note         = {Machine review of arXiv:2502.06515}
}
abstract

Exploring measures to improve financial networks and mitigate systemic risks is an ongoing challenge. We study claims trading, a notion defined in Chapter 11 of the U.S. Bankruptcy Code. For a bank $v$ in distress and a trading partner $w$, the latter is taking over some claims of $v$ and in return giving liquidity to $v$. The idea is to rescue $v$ (or mitigate contagion effects from $v$'s insolvency). We focus on the impact of trading claims fractionally, when $v$ and $w$ can agree to trade only part of a claim. In addition, we study donations, in which $w$ only provides liquidity to $v$. They can be seen as special claims trades. When trading a single claim or making a single donation in networks without default cost, we show that it is impossible to strictly improve the assets of both banks $v$ and $w$. Since the goal is to rescue $v$ in distress, we study creditor-positive trades, in which $v$ improves and $w$ remains indifferent. We show that an optimal creditor-positive trade that maximizes the assets of $v$ can be computed in polynomial time. It also yields a (weak) Pareto-improvement for all banks in the entire network. In networks with default cost, we obtain a trade in polynomial time that weakly Pareto-improves all assets over the ones resulting from the optimal creditor-positive trade. We generalize these results to trading multiple claims for which $v$ is the creditor. Instead, when trading claims with a common debtor $u$, we obtain NP-hardness results for computing trades in networks with default cost that maximize the assets of the creditors and Pareto-improve the assets in the network. Similar results apply when $w$ donates to multiple banks in networks with default costs. For networks without default cost, we give an efficient algorithm to compute optimal donations to multiple banks.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 25 canonical work pages

  1. [1]

    Strategic payments in financial networks

    Nils Bertschinger, Martin Hoefer, and Daniel Schmand. Strategic payments in financial networks. In Proc.\ 11th Symp.\ Innov.\ Theoret.\ Comput.\ Sci.\ (ITCS) , pages 46:1--46:16, 2020

    work page 2020

  2. [2]

    Decentralized clearing in financial networks

    P \'e ter Cs \'o ka and Jean-Jacques Herings. Decentralized clearing in financial networks. Manag.\ Sci. , 64(10):4681--4699, 2018

    work page 2018

  3. [3]

    Uniqueness of clearing payment matrices in financial networks

    P \'e ter Cs \'o ka and Jean-Jacques Herings. Uniqueness of clearing payment matrices in financial networks. Math.\ Oper.\ Res. , 49(1):232--250, 2024

    work page 2024

  4. [4]

    What is the price for lending in financial networks? In Proc.\ 25th Int.\ Conf.\ Princ.\ Pract.\ Multi-Agent Syst

    B \'e ni Egressy, Andreas Plesner, and Roger Wattenhofer. What is the price for lending in financial networks? In Proc.\ 25th Int.\ Conf.\ Princ.\ Pract.\ Multi-Agent Syst. (PRIMA) , pages 120--135, 2024

    work page 2024

  5. [5]

    Systemic risk in financial systems

    Larry Eisenberg and Thomas Noe. Systemic risk in financial systems. Manag.\ Sci. , 47(2):236--249, 2001

    work page 2001

  6. [6]

    The complexity of debt swapping

    Henri Froese, Martin Hoefer, and Lisa Wilhelmi. The complexity of debt swapping. CoRR , abs/2302.11250, 2023

    work page arXiv 2023

  7. [7]

    Algorithms for claims trading

    Martin Hoefer, Carmine Ventre, and Lisa Wilhelmi. Algorithms for claims trading. In Proc.\ 41st Symp.\ Theoret.\ Aspects Comput.\ Sci.\ (STACS) , pages 42:1--42:17, 2024

    work page 2024

  8. [8]

    Seniorities and minimal clearing in financial network games

    Martin Hoefer and Lisa Wilhelmi. Seniorities and minimal clearing in financial network games. In Proc.\ 15th Symp.\ Algorithmic Game Theory (SAGT) , pages 187--204, 2022

    work page 2022

Show all 26 references
  1. [9]

    Strong approximations and irrationality in financial networks with derivatives

    Stavros Ioannidis, Bart de Keijzer, and Carmine Ventre. Strong approximations and irrationality in financial networks with derivatives. In Proc.\ 49th Int.\ Colloq.\ Autom.\ Lang.\ Programming (ICALP) , pages 76:1--76:18, 2022

    2022

  2. [10]

    Clearing financial networks with derivatives: From intractability to algorithms

    Stavros Ioannidis, Bart de Keijzer, and Carmine Ventre. Clearing financial networks with derivatives: From intractability to algorithms. CoRR , abs/2312.05139, 2023

    2023 arXiv

  3. [11]

    Financial networks with singleton liability priorities

    Stavros Ioannidis, Bart de Keijzer, and Carmine Ventre. Financial networks with singleton liability priorities. Theor.\ Comput.\ Sci. , 963:113965, 2023

    2023

  4. [12]

    Financial network games

    Panagiotis Kanellopoulos, Maria Kyropoulou, and Hao Zhou. Financial network games. In Proc.\ 2nd Int.\ Conf.\ AI in Finance (ICAIF) , pages 26:1--26:9, 2021

    2021

  5. [13]

    Forgiving debt in financial network games

    Panagiotis Kanellopoulos, Maria Kyropoulou, and Hao Zhou. Forgiving debt in financial network games. In Proc.\ 31st Int.\ Joint Conf.\ Artif.\ Intell.\ (IJCAI) , pages 335--341, 2022

    2022

  6. [14]

    Debt transfers in financial networks: Complexity and equilibria

    Panagiotis Kanellopoulos, Maria Kyropoulou, and Hao Zhou. Debt transfers in financial networks: Complexity and equilibria. In Proc.\ 22nd Conf.\ Auton.\ Agents and Multi-Agent Syst.\ (AAMAS) , pages 260--268, 2023

    2023

  7. [15]

    Dynamic interventions for networked contagions

    Marios Papachristou, Siddhartha Banerjee, and Jon Kleinberg. Dynamic interventions for networked contagions. In Proc.\ 32nd World Wide Web Conf.\ (WWW) , pages 3519--3529, 2023

    2023

  8. [16]

    Allocating stimulus checks in times of crisis

    Marios Papachristou and Jon Kleinberg. Allocating stimulus checks in times of crisis. In Proc.\ 31st World Wide Web Conf.\ (WWW) , pages 16--26, 2022

    2022

  9. [17]

    Network-aware strategies in financial systems

    P \' a l Andr \' a s Papp and Roger Wattenhofer. Network-aware strategies in financial systems. In Proc.\ 47th Int.\ Colloq.\ Autom.\ Lang.\ Programming (ICALP) , pages 91:1--91:17, 2020

    2020

  10. [18]

    Debt swapping for risk mitigation in financial networks

    P \'a l Andr \'a s Papp and Roger Wattenhofer. Debt swapping for risk mitigation in financial networks. In Proc.\ 22nd Conf.\ Econ.\ Comput.\ (EC) , pages 765--784, 2021

    2021

  11. [19]

    Default ambiguity: Finding the best solution to the clearing problem

    P \' a l Andr \' a s Papp and Roger Wattenhofer. Default ambiguity: Finding the best solution to the clearing problem. In Proc.\ 17th Conf.\ Web and Internet Econ.\ (WINE) , pages 391--409, 2021

    2021

  12. [20]

    Sequential defaulting in financial networks

    P \' a l Andr \' a s Papp and Roger Wattenhofer. Sequential defaulting in financial networks. In Proc.\ 12th Symp.\ Innov.\ Theoret.\ Comput.\ Sci.\ (ITCS) , pages 52:1--52:20, 2021

    2021

  13. [21]

    Failure and rescue in an interbank network

    Leonard Rogers and Luitgard Veraart. Failure and rescue in an interbank network. Manag.\ Sci. , 59(4):882--898, 2013

    2013

  14. [22]

    Finding clearing payments in financial networks with credit default swaps is PPAD -complete

    Steffen Schuldenzucker, Sven Seuken, and Stefano Battiston. Finding clearing payments in financial networks with credit default swaps is PPAD -complete. In Proc.\ 8th Symp.\ Innov.\ Theoret.\ Comput.\ Sci.\ (ITCS) , pages 32:1--32:20, 2017

    2017

  15. [23]

    Default ambiguity: Credit default swaps create new systemic risks in financial networks

    Steffen Schuldenzucker, Sven Seuken, and Stefano Battiston. Default ambiguity: Credit default swaps create new systemic risks in financial networks. Manag.\ Sci. , 66(5):1981--1998, 2020

    1981

  16. [24]

    Selfishly cancelling debts can reduce systemic risk

    Jinyun Tong, Bart de Keijzer, and Carmine Ventre. Selfishly cancelling debts can reduce systemic risk. In Proc.\ 27th European Conf.\ Aritf.\ Intell.\ (ECAI) , pages 3397--3404, 2024

    2024

  17. [25]

    Reducing systemic risk in financial networks through donations

    Jinyun Tong, Bart De Keijzer, and Carmine Ventre. Reducing systemic risk in financial networks through donations. In Proc.\ 27th European Conf.\ Aritf.\ Intell.\ (ECAI) , pages 3405--3412, 2024

    2024

  18. [26]

    A strategic analysis of prepayments in financial credit networks

    Hao Zhou, Yongzhao Wang, Konstantinos Varsos, Nicholas Bishop, Rahul Savani, Anisoara Calinescu, and Michael Wooldridge. A strategic analysis of prepayments in financial credit networks. In Proc.\ 33rd Int.\ Joint Conf.\ Artif.\ Intell.\ (IJCAI) , pages 3040--3048, 2024

    2024

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.