arxiv: 2605.12508 · v1 · submitted 2026-03-23 · 💻 cs.SI · cs.CR· q-fin.RM

Recognition: no theorem link

Interoperability Effects: Extending DeFi Lending Risk Models to Multi-Chain Environments

Hasret Ozan Sevim

Authors on Pith no claims yet

Pith reviewed 2026-05-15 00:10 UTC · model grok-4.3

classification 💻 cs.SI cs.CRq-fin.RM

keywords DeFilending protocolsmulti-chaincross-chain bridgesTVLrevenueinteroperabilityrisk models

0 comments

The pith

Cross-chain bridge activity significantly affects DeFi lending TVL and revenue with heterogeneous directions.

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

This paper investigates the impact of cross-chain interoperability on decentralized lending protocols in a multi-chain DeFi environment. It uses empirical models on data spanning 15 protocols and 53 bridges across nine EVM chains from late 2022 to early 2025 to show how bridge-related metrics influence total value locked and revenue. Bridge volume stands out as a key factor with varying effects by chain category, while more bridge integrations tend to reduce both TVL and revenue, pointing to liquidity leaving the protocols. The findings highlight the need to update risk management approaches to include these cross-chain elements and differentiate between layers like Ethereum and its layer-2s.

Core claim

Bridge volume exerts a significant effect on TVL and revenue across different categories, though the direction varies heterogeneously. Increased bridge integrations are associated with decreased TVL and protocol revenue across categories, indicating liquidity escapes from those lending ecosystems. Liquidations produce heterogeneous effects across categories. New network launches do not have as significant relationships with TVL and revenue while bridge hacks show a significant and positive relationship. Ethereum attracts large depositors, while layer-2s skew toward retail participation.

What carries the argument

Panel regression fixed effects and OLS models using bridge volume, number of integrations, liquidations, and hacks as explanatory variables for TVL and revenue performance.

If this is right

Risk models for DeFi lending protocols must incorporate cross-chain metrics to accurately assess performance.
A layer-aware approach is necessary to reflect differences between Ethereum, alternative L1s, and L2 networks.
Liquidity can migrate away from lending protocols through increased bridge integrations.
Bridge hacks are positively associated with performance metrics, possibly due to increased activity or scrutiny.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Protocol designers could strategically limit or manage bridge integrations to retain liquidity within their ecosystems.
Users and analysts might track bridge volume as a leading indicator for protocol health in multi-chain setups.
Extending this analysis to non-EVM chains could reveal if the patterns hold more broadly.

Load-bearing premise

The observational data allows interpreting bridge activity as causally affecting protocol TVL and revenue without substantial endogeneity or bias from unmeasured market conditions.

What would settle it

If adding controls for overall cryptocurrency market conditions and other DeFi metrics eliminates the significant effects of bridge volume and integrations, the causal interpretation would be challenged.

Figures

Figures reproduced from arXiv: 2605.12508 by Hasret Ozan Sevim.

**Figure 2.** Figure 2: Cross-Blockchain Bridge Volume by Network Groups [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 5.** Figure 5: Deposit Volume in the Lending Protocols by Network Groups [PITH_FULL_IMAGE:figures/full_fig_p003_5.png] view at source ↗

**Figure 4.** Figure 4: Active User Count in Lending Protocols by Network Groups [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗

read the original abstract

On-chain lending has expanded across multiple distributed ledgers as DeFi becomes increasingly multi-chain. This environment introduces novel technical and financial mechanisms, particularly cross-blockchain communication and asset transfer protocols, yet cross-chain elements remain understudied in lending protocol risk management. To address this gap, we applied panel regression fixed effects and OLS models to empirically analyze cross-blockchain interoperability solutions, using TVL and total revenue as performance proxies from October 2022 to January 2025. Our data set covers 15 decentralized lending protocols and 53 cross-chain bridges across 9 EVM-compatible blockchains, categorized as Ethereum, alternative layer-1s, and Ethereum layer-2 networks. Results reveal that cross-chain activity impacts on protocol performance. Bridge volume emerges as a critical driver, exerts a significant effect on TVL and revenue across different categories, though the direction of this effect varies heterogeneously. Increased bridge integrations are associated with decreased TVL and protocol revenue across categories, indicating liquidity escapes from those lending ecosystems. Liquidations produce heterogeneous effects across categories. New network launches do not have as significant relationships with TVL and revenue while bridge hacks show a significant and positive relationship. High R-squared values confirm meaningful explanatory power. We further show Ethereum attracts large depositors, while layer-2s skew toward retail participation. We conclude that effective DeFi risk models should incorporate cross-chain metrics and adopt a layer-aware approach to accurately reflect the evolving multi-chain landscape.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Bridge integrations correlate with lower TVL and revenue in this multi-chain lending sample, with some layer-specific patterns, but the regressions leave endogeneity unaddressed so causal claims stay tentative.

read the letter

The paper pulls together data on 15 lending protocols and 53 bridges across Ethereum, other L1s, and L2s from October 2022 through January 2025, then runs panel fixed-effects and OLS models on TVL and revenue. It reports that higher bridge integrations tie to drops in both metrics, with direction varying by category, while hacks link positively and new network launches show little relation. It also flags that Ethereum pulls larger depositors while L2s skew retail. That combination of bridge metrics with lending outcomes across layers is not something the cited single-chain work already covered, so the empirical patterns are the main addition. The standard regression tools fit the observational data they have, and the high R-squared values indicate the variables capture real variation. The layer participation split is a concrete detail that could feed into risk monitoring. The soft spot is identification. Fixed effects absorb protocol-level differences but do not handle time-varying factors like overall market TVL shifts, ETH volatility, or competing yields that could move both bridge activity and protocol performance together. The abstract gives no sign of extra controls, lags, or instruments, so reading the negative coefficients as liquidity escapes is a stretch. This is for DeFi risk researchers or protocol teams who want empirical hooks on cross-chain flows rather than a finished causal model. The data collection and basic setup are solid enough that a serious referee should see it, with requests for robustness and identification checks.

Referee Report

3 major / 2 minor

Summary. The paper empirically examines the impact of cross-chain interoperability on DeFi lending protocols using panel fixed-effects and OLS models on data from 15 protocols and 53 bridges across 9 EVM chains from October 2022 to January 2025. It claims that bridge volume has heterogeneous significant effects on TVL and revenue, while increased bridge integrations are associated with decreased TVL and protocol revenue, interpreted as liquidity escapes. Additional findings include heterogeneous effects from liquidations, positive effects from bridge hacks, and differences in user participation between Ethereum and layer-2 networks. The models report high R-squared values, supporting the need to incorporate cross-chain metrics in risk models.

Significance. If the results hold after addressing potential biases, this study contributes to extending DeFi risk models to multi-chain settings by providing evidence on how interoperability mechanisms influence protocol performance metrics like TVL and revenue. It highlights risks such as liquidity outflows associated with bridge integrations and suggests layer-aware approaches. The use of on-chain data from multiple chains and protocols is a strength, offering insights into an understudied aspect of DeFi.

major comments (3)

[Abstract] Abstract and results: The central claim interprets negative associations between bridge integrations and TVL/revenue as 'liquidity escapes' from lending ecosystems, but the fixed-effects and OLS specifications do not address time-varying endogeneity (e.g., simultaneous decisions on bridges and lending, or omitted factors like ETH volatility and competing yields), undermining the causal interpretation of interoperability effects.
[Empirical Analysis] Empirical strategy: No robustness checks, instrumental variables, lagged instruments, or explicit market controls are described to support conditional exogeneity of bridge metrics after protocol fixed effects; this is load-bearing for the heterogeneous directional claims across Ethereum, alt-L1, and L2 categories.
[Results] Results reporting: High R-squared is noted but without p-values, standard errors, or discussion of selection/reverse causality, the evidence for significant bridge-volume effects and liquidation heterogeneity cannot be fully evaluated.

minor comments (2)

[Data] Data section: Provide precise definitions and sources for 'bridge integrations' and 'bridge volume' variables, including aggregation across the 53 bridges and 9 chains.
[Conclusion] Discussion: The layer-aware recommendation could include a concrete example of how to adjust existing risk models (e.g., incorporating bridge-volume terms into liquidation thresholds).

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive comments, which highlight important areas for strengthening the empirical identification and reporting in our manuscript. We address each major comment below and indicate the planned revisions.

read point-by-point responses

Referee: [Abstract] Abstract and results: The central claim interprets negative associations between bridge integrations and TVL/revenue as 'liquidity escapes' from lending ecosystems, but the fixed-effects and OLS specifications do not address time-varying endogeneity (e.g., simultaneous decisions on bridges and lending, or omitted factors like ETH volatility and competing yields), undermining the causal interpretation of interoperability effects.

Authors: We agree that the phrasing 'liquidity escapes' implies a stronger causal interpretation than the fixed-effects and OLS models can support, given potential time-varying endogeneity from simultaneous bridge and lending decisions or omitted market factors. In the revised manuscript, we will replace this language in the abstract and results with more neutral terms such as 'associated with liquidity outflows' and add an explicit limitations paragraph discussing endogeneity concerns. We will also incorporate additional controls for ETH volatility and competing yields to the extent permitted by available data. revision: partial
Referee: [Empirical Analysis] Empirical strategy: No robustness checks, instrumental variables, lagged instruments, or explicit market controls are described to support conditional exogeneity of bridge metrics after protocol fixed effects; this is load-bearing for the heterogeneous directional claims across Ethereum, alt-L1, and L2 categories.

Authors: We acknowledge that the current draft lacks explicit robustness checks and market controls. In the revision, we will add lagged bridge volume and integration variables to address potential simultaneity, include controls for ETH price volatility and aggregate DeFi yields, and report robustness checks including alternative specifications and category-specific subsample analyses. While suitable instruments for IV estimation are not readily available in this on-chain setting, these additions will provide stronger support for the conditional exogeneity assumption underlying the heterogeneous results. revision: yes
Referee: [Results] Results reporting: High R-squared is noted but without p-values, standard errors, or discussion of selection/reverse causality, the evidence for significant bridge-volume effects and liquidation heterogeneity cannot be fully evaluated.

Authors: The full manuscript tables include standard errors and p-values, but we recognize that the main text and abstract do not sufficiently highlight them or discuss selection and reverse causality. We will expand the results section to explicitly report and interpret statistical significance for key coefficients, add a dedicated paragraph addressing potential selection bias and reverse causality, and clarify how protocol fixed effects and the planned robustness checks help mitigate these issues. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical regression on external on-chain data

full rationale

The paper applies standard panel fixed-effects and OLS regressions to observed TVL, revenue, bridge volume, and integration counts drawn from external blockchain data (Oct 2022–Jan 2025). Coefficients are estimated directly from these inputs; no fitted parameter is subsequently reused to generate a “prediction” that is then compared back to the same data, and no self-citation supplies a load-bearing uniqueness theorem or ansatz. The central claims are therefore statistical associations, not derivations that reduce to their own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard econometric assumptions rather than new axioms or invented entities; no free parameters beyond the usual regression coefficients are introduced.

axioms (2)

domain assumption Linear relationship between bridge metrics and TVL/revenue after controlling for fixed effects
Invoked by the choice of OLS and panel regression models in the abstract.
domain assumption No substantial omitted variable bias from unmeasured market-wide shocks
Required for interpreting coefficients as meaningful effects.

pith-pipeline@v0.9.0 · 5559 in / 1415 out tokens · 44212 ms · 2026-05-15T00:10:12.680809+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

39 extracted references · 39 canonical work pages

[1]

Decentralized finance: On blockchain- and smart contract-based financial markets,

F. Sch ¨ar, “Decentralized finance: On blockchain- and smart contract-based financial markets,” FRB of St. Louis Review, Apr. 2021. URL: https: //dx.doi.org/10.20955/r.103.153-74

work page doi:10.20955/r.103.153-74 2021
[2]

In: Proc

C. G. Harris, “Cross-chain technologies: Challenges and opportunities for blockchain interoperability,” in 2023 IEEE International Conference on Omni-layer Intelligent Systems (COINS), Berlin, Germany, 2023, pp. 1–6. URL: https://doi.org/10.1109/COINS57856.2023.10189298

work page doi:10.1109/coins57856.2023.10189298 2023
[3]

1, 2025)

DeFiLlama, DeFi analytics platform, URL: https://defillama.com (ac- cessed Jan. 1, 2025)

work page 2025
[4]

arXiv:2404.11745 (2025)

Y . Luo, Y . Feng, J. Xu, and P. Tasca, “Piercing the veil of TVL: DeFi reappraised,” arXiv preprint arXiv:2404.11745, 2025. URL: https://arxiv. org/abs/2404.11745

work page arXiv 2025
[5]

Automated risk management mechanisms in DeFi lending protocols: A cross-chain comparative anal- ysis of Aave and Compound,

E. Iftikhar, W. Wei, and J. Cartlidge, “Automated risk management mechanisms in DeFi lending protocols: A cross-chain comparative anal- ysis of Aave and Compound,” arXiv preprint arXiv:2506.12855. URL: https://doi.org/10.48550/arXiv.2506.12855

work page doi:10.48550/arxiv.2506.12855
[6]

Systemic fragility in decentralized markets,

A. Lehar and C. A. Parlour, “Systemic fragility in decentralized markets,” BIS Working Papers No. 1062, Bank for International Settlements, 2022. URL: https://www.bis.org/publ/work1062.pdf (accessed Jan. 1, 2025)

work page 2022
[7]

Time-varying bidirectional causal relationships between transaction fees and economic activity of subsystems utilizing the Ethereum blockchain network,

L. Ante and A. Saggu, “Time-varying bidirectional causal relationships between transaction fees and economic activity of subsystems utilizing the Ethereum blockchain network,” J. Risk Financial Manag., vol. 17, p. 19, 2024. URL: https://doi.org/10.3390/jrfm17010019

work page doi:10.3390/jrfm17010019 2024
[8]

Measuring DeFi risk,

J. Bertomeu, X. Martin, and I. Sall, “Measuring DeFi risk,” Finance Research Letters, vol. 63, 105321, 2024. URL: https://doi.org/10.1016/ j.frl.2024.105321

work page arXiv 2024
[9]

An empirical study of DeFi liquidations: Incentives, risks, and instabilities,

K. Qin, L. Zhou, P. Gamito, P. Jovanovic, and A. Gervais, “An empirical study of DeFi liquidations: Incentives, risks, and instabilities,” in Proc. 21st ACM Internet Measurement Conference (IMC ’21), 2021, pp. 336–

work page 2021
[10]

URL: https://doi.org/10.1145/3487552.3487811

work page doi:10.1145/3487552.3487811
[11]

Toxic liquidation spirals,

J. Warmuz, A. Chaudhary, and D. Pinna, “Toxic liquidation spirals,” arXiv preprint arXiv:2212.07306, 2022. URL: https://doi.org/10.48550/ arXiv.2212.07306

work page arXiv 2022
[13]

URL: https://arxiv.org/abs/2206.11973

work page arXiv
[14]

Decentralized lending and its users: Insights from Compound,

K. Saengchote, “Decentralized lending and its users: Insights from Compound,” arXiv preprint arXiv:2212.05734, 2022. URL: https://arxiv. org/abs/2212.05734

work page arXiv 2022
[15]

Total value locked volatility in DeFi: Empirical evidence from market crashes,

F. Mottaghi and B. I. Steininger, “Total value locked volatility in DeFi: Empirical evidence from market crashes,” SSRN, Feb. 2025. URL: http: //dx.doi.org/10.2139/ssrn.5161601

work page doi:10.2139/ssrn.5161601 2025
[16]

Simulation of dynamic performance of DeFi protocol based on historical crypto market behavior,

I. Grigorova, A. Karamfilov, R. Merakov, and A. Efremov, “Simulation of dynamic performance of DeFi protocol based on historical crypto market behavior,” Risks, vol. 12, no. 1, 2024. URL: https://doi.org/10. 3390/risks12010003

work page 2024
[17]

Decentralized finance (DeFi) projects: A study of key performance indicators in terms of DeFi protocols’ valuations,

D. Metelski and J. Sobieraj, “Decentralized finance (DeFi) projects: A study of key performance indicators in terms of DeFi protocols’ valuations,” Int. J. Financial Stud., vol. 10, no. 4, p. 108, 2022. URL: https://doi.org/10.3390/ijfs10040108

work page doi:10.3390/ijfs10040108 2022
[18]

The Graph Explorer,

The Graph, “The Graph Explorer,” URL: https://thegraph.com/ explorer/profile/0x7e8f317a45d67e27e095436d2e0d47171e7c769f?view= Subgraphs&chain=arbitrum-one (accessed Jan. 1, 2025)

work page 2025
[19]

Ethereum Block Explorer,

Etherscan, “Ethereum Block Explorer,” URL: https://etherscan.io (ac- cessed Jan. 1, 2025)

work page 2025
[20]

Arbitrum Block Explorer,

Arbiscan, “Arbitrum Block Explorer,” URL: https://arbiscan.io (accessed Jan. 1, 2025)

work page 2025
[21]

Base Block Explorer,

Basescan, “Base Block Explorer,” URL: https://basescan.org (accessed Jan. 1, 2025)

work page 2025
[22]

1, 2025)

zkSync Explorer, URL: https://explorer.zksync.io (accessed Jan. 1, 2025)

work page 2025
[23]

1, 2025)

OP Mainnet Explorer, URL: https://optimistic.etherscan.io (accessed Jan. 1, 2025)

work page 2025
[24]

1, 2025)

Gnosis Chain Explorer, URL: https://gnosisscan.io (accessed Jan. 1, 2025)

work page 2025
[25]

Avalanche Block Explorer,

SnowTrace, “Avalanche Block Explorer,” URL: https://snowtrace.io (ac- cessed Jan. 1, 2025)

work page 2025
[26]

BNB Smart Chain Explorer,

BscScan, “BNB Smart Chain Explorer,” URL: https://bscscan.com (ac- cessed Jan. 1, 2025)

work page 2025
[27]

Polygon Block Explorer,

PolygonScan, “Polygon Block Explorer,” URL: https://polygonscan.com (accessed Jan. 1, 2025)

work page 2025
[28]

Crypto Fear & Greed Index,

Alternative.me, “Crypto Fear & Greed Index,” URL: https://alternative. me/crypto/fear-and-greed-index/ (accessed Jan. 1, 2025)

work page 2025
[29]

Modularizing decentralised finance: The DeFi layered stack revisited,

E. Rossi, “Modularizing decentralised finance: The DeFi layered stack revisited,” SSRN, Jan. 2023. URL: https://dx.doi.org/10.2139/ssrn. 4365222

work page doi:10.2139/ssrn 2023
[30]

A survey of layer- two blockchain protocols,

A. Gangwal, H. R. Gangavalli, and A. Thirupathi, “A survey of layer- two blockchain protocols,” arXiv preprint arXiv:2204.08032, 2022. URL: https://arxiv.org/abs/2204.08032

work page arXiv 2022
[31]

Liquidity frag- mentation or optimization? Analyzing automated market makers across Ethereum and rollups,

K. Gogol, M. Schneider, C. Tessone, and B. Livshits, “Liquidity frag- mentation or optimization? Analyzing automated market makers across Ethereum and rollups,” arXiv preprint arXiv:2410.10324, 2025. URL: https://arxiv.org/abs/2410.10324

work page arXiv 2025
[32]

arXiv:2401.16353 (2024)

K. Gogol, B. Kraner, M. Schlosser, T. Yan, C. Tessone, and B. Stiller, “Empirical and theoretical analysis of liquid staking protocols,” arXiv preprint arXiv:2401.16353, 2024. URL: https://arxiv.org/abs/2401.16353

work page arXiv 2024
[33]

An analysis of fixed-spread liquidation lending in DeFi,

C. Moallemi and U. Patange, “An analysis of fixed-spread liquidation lending in DeFi,” in Financial Cryptography and Data Security. FC 2024 International Workshops, ser. Lecture Notes in Computer Science. Cham: Springer, 2025, vol. 14746, pp. 105–127. URL: https://doi.org/10.1007/ 978-3-031-69231-4 8 TABLE VI VARIANCEINFLATIONFACTOR(VIF) RESULTS Variable ...

work page 2024
[34]

Arbitrage in crypto markets: An analysis of primary Ethereum blockchain data,

M. Hansson, “Arbitrage in crypto markets: An analysis of primary Ethereum blockchain data,” SSRN, 2022. URL: https://dx.doi.org/10. 2139/ssrn.4278272

work page 2022
[35]

SSRN (2024)

C. Bertucci, L. Bertucci, M. Gontier Delaunay, O. Gu ´eant, and M. Lesbre, “Agents’ behavior and interest rate model optimization in DeFi lending,” 2024. URL: https://doi.org/10.2139/ssrn.4802776

work page doi:10.2139/ssrn.4802776 2024
[36]

DeFi leverage,

L. Heimbach and W. Huang, “DeFi leverage,” BIS Working Papers, Bank for International Settlements, no. 1171, Mar. 2024. URL: https: //ideas.repec.org/p/bis/biswps/1171.html (accessed Mar. 10, 2026)

work page 2024
[37]

An analysis of the market risk to participants in the Compound protocol,

H.-T. Kao, T. Chitra, R. Chiang, J. Morrow, and Gauntlet, “An analysis of the market risk to participants in the Compound protocol,” 2020. URL: https://api.semanticscholar.org/CorpusID:231619656

work page 2020
[38]

J. M. Wooldridge,Econometric Analysis of Cross Section and Panel Data, Cambridge, MA, USA: MIT Press, 2010

work page 2010
[39]

Decentralized finance protocols: Evaluating collateral- ization mechanisms,

I. Sinyugin, “Decentralized finance protocols: Evaluating collateral- ization mechanisms,” B.Sc. thesis, University of Twente, 2023. URL: https://essay.utwente.nl/95990/

work page 2023
[40]

The paradox of adversarial liquidation in decentralised lending,

S. N. Cohen, M. Sabate-Vidales, L. Szpruch, and M. G. Delaunay, “The paradox of adversarial liquidation in decentralised lending,” 2023. URL: https://doi.org/10.2139/ssrn.4540333

work page doi:10.2139/ssrn.4540333 2023