{"paper_id":"2611.20589","title":"Beyond Marginal Risk: ESG and Multidimensional Fragility in Equity Markets","authors":["Minxuan Hu","Jiayu Yi","Ziheng Chen","Wenxi Sun","Qishi Zhan"],"paper_date":"2026-11-16","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2611.20589","page_url":"https://thequant.space/flowcharts/beyond-marginal-risk-esg-and-multidimensional-fragility-in-equity-markets/","scores":{"empirical_rigor":8,"math_complexity":6.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes to test whether apparent ESG resilience survives explicit conventional risk adjustment, to locate any remaining ESG association within overlapping adverse states, and to assess whether that pattern depends on the MSCI score field used to represent ESG.","quote":"We first test whether apparent ESG resilience survives explicit conventional risk adjustment, then examine whether any remaining coefficient pattern is concentrated within overlapping adverse states, and finally assess whether that pattern depends on the MSCI score field used to represent ESG.","summary":"Do ESG characteristics provide financial resilience beyond conventional risk, and is that inference stable to the financial state and ESG score being measured? We study a point-in-time panel of nonfinancial S&P 500 constituents from January 2019 through November 2025. Once ex ante return…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The empirical design is point‑in‑time: ESG and other explanatory variables are observed strictly before the outcome month, preventing contemporaneous outcomes from entering the regressors.","type":"data","stated":"explicit","quote":"These rules make the empirical design point-in-time and prevent contemporaneous outcomes from entering the explanatory variables.","support":"full"},{"assumption":"Conventional risk exposures (beta, prior volatility, prior liquidity) are sufficient to capture the relevant financial risk dimensions.","type":"model","stated":"explicit","quote":"Conventional risk adjustment is central to that distinction.","support":"full"},{"assumption":"Adverse market states are defined by the 80th‑percentile tails of loss, realized volatility, and CHL spread.","type":"statistical","stated":"explicit","quote":"The loss and RVOL thresholds, controls, ex ante exposure blocks, firm and month fixed effects, and covariance estimator remain unchanged.","support":"partial"},{"assumption":"Wild‑cluster bootstrap with two‑way firm and month clustering provides valid inference given the limited number of months.","type":"statistical","stated":"explicit","quote":"Because the state variables vary at the monthly level and the sample contains 83 calendar months, we additionally use a wild-cluster-bootstrap procedure with the exact OLS specifications, two-way firm and month clustering, Rademacher weights, the null-imposed FNW11 procedure, and 9,999 replications; the bootstrap clustering dimension is the smaller month dimension.","support":"full"},{"assumption":"The sample of non‑financial S&P 500 constituents with linked market, ESG, and accounting data is representative of the observable linked constituent population.","type":"data","stated":"explicit","quote":"The linked sources become substantially less complete near some exits, especially M&A-related exits, so the regression estimand should be understood as applying to the observable linked constituent sample rather than to all firms approaching index exit.","support":"partial"},{"assumption":"Trading‑cost measurement via the close‑high‑low (CHL) spread accurately reflects liquidity pressure in the absence of intraday quote data.","type":"other","stated":"explicit","quote":"Trading cost is measured with the close-high-low bid-ask spread estimator of Abdi and Ranaldo [2017], which is designed for settings without intraday quote data. We aggregate valid within-month estimates and use the logarithm of the monthly CHL spread.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Compustat market and accounting data","provider":"S&P Global Market Intelligence","kind":"fundamentals","access":"commercial","period":null,"universe":"non‑financial S&P 500 constituents","quote":"Daily market data and annual accounting variables are obtained from Compustat [S&P Global Market Intelligence, 2026].","support":"full","known_id":"compustat"},{"name":"MSCI ESG monthly snapshots","provider":"MSCI","kind":"alternative","access":"commercial","period":null,"universe":"non‑financial S&P 500 constituents","quote":"ESG variables come from monthly MSCI snapshots [MSCI, 2026].","support":"full"},{"name":"S&P 500 constituent membership records","provider":"Wikipedia contributors","kind":"benchmark","access":"public","period":null,"universe":"S&P 500","quote":"Historical constituent records [Wikipedia contributors, 2026] form the universe to which market, ESG, and accounting observations are linked.","support":"full"}],"known":[{"id":"compustat","name":"Compustat","provider":"S&P Global","kind":"fundamentals","access":"commercial","quote":"Daily market data and annual accounting variables are obtained from Compustat [S&P Global Market Intelligence, 2026].","mentions":2,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"After explicitly modeling ex‑ante return, volatility, and liquidity exposures, the broad downside‑resilience pattern attenuates and none of the three primary marginal ESG×Down tests survives Holm adjustment under either conventional or wild‑cluster‑bootstrap i","kind":"statistical","quote":"Once ex ante return, volatility, and liquidity exposures are modeled explicitly, the broad downside-resilience pattern attenuates and none of the three primary marginal ESG $\\times$ Down tests survives Holm adjustment under either conventional or wild-cluster-bootstrap inference.","support":"full"},{"finding":"An exact decomposition shows that the negative coefficient on the inclusive Return+RVOL state is almost entirely allocated to observations that also breach the CHL trading‑cost tail, while the mutually exclusive Return+RVOL‑only component is essentially zero.","kind":"statistical","quote":"A post-primary exact decomposition then shows that the negative Return+RVOL coefficient is almost entirely allocated to observations that also breach the CHL trading-cost tail, while the mutually exclusive Return+RVOL-only component is essentially zero.","support":"full"},{"finding":"On the identical sample, standardized IAS and WAKIS scores imply materially different fitted associations in adverse market states; at a 5 % market decline the difference is about one probability percentage point per score standard deviation.","kind":"statistical","quote":"Finally, a paired standardized comparison on an identical sample shows that MSCI's downstream Industry Adjusted Score and upstream Weighted Average Key Issue Score imply materially different fitted associations in adverse market states: at a 5% market decline, the difference is about one probability percentage point per score standard deviation.","support":"full"},{"finding":"In the fully risk‑adjusted marginal specification, the Industry Adjusted ESG×Down coefficient in the return‑loss model is –0.0186 (raw p = 0.243, Holm‑adjusted p = 0.420).","kind":"statistical","quote":"In the return-loss model, the coefficient is −0.0186 with a raw p-value of 0.243 and a Holm-adjusted p-value of 0.420.","support":"full"},{"finding":"When conventional exposures are omitted, ESG×Down is –0.1088 (p < 0.001); adding beta, prior volatility, or prior liquidity individually reduces the magnitude, and with all three blocks the coefficient becomes –0.0702, attenuating the effect by about 35 %.","kind":"statistical","quote":"Without beta, prior volatility, or prior liquidity, ESG × Down is −0.1088 (p < 0.001), ESG × Up is −0.0225, and their difference is significant (p = 0.0028). Adding beta alone reduces the Down coefficient to −0.0826; prior volatility alone yields −0.0730; and prior liquidity alone yields −0.0716.","support":"partial"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":16,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":3,"extracted_at":"2026-11-17T15:29:18Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}