{"paper_id":"2410.14984","title":"Risk Aggregation and Allocation in the Presence of Systematic Risk via Stable Laws","authors":["Andrew Fleck","Edward Furman","Yang Shen"],"paper_date":"2024-10-19","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2410.14984v1","page_url":"https://thequant.space/flowcharts/risk-aggregation-and-allocation-in-the-presence-of-systemati/","scores":{"empirical_rigor":1,"math_complexity":9.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a modification to the classical individual risk model by using multivariate Stable distributions to model company‑wide losses and computes the Tail Conditional Expectation of aggregate risks together with the corresponding allocations.","quote":"We propose a modification to the classical individual risk model that allows us to model company-wide losses via the class of Multivariate Stable Distributions. Stable Distributions incorporate many of the unpleasant features required for a realistic risk model while maintaining tractable aggregation and dependence results.","summary":"In order to properly manage risk, practitioners must understand the aggregate risks they are exposed to. Additionally, to properly price policies and calculate bonuses the relative riskiness of individual business units must be well understood. Certainly, Insurers and Financiers are interested in…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The location parameter µ is set to zero in the construction of the Stable vector.","type":"model","stated":"explicit","quote":"Assume µ = 0 and define Z(j) ∼Sα(1, 1, 0) i.i.d. and X :=    Pn j=1 sjγ1/α j Z(j) α̸ = 1 Pn j=1 sjγj(Z(j) + 2 π ln γj) α = 1.","support":"full"},{"assumption":"The analysis focuses on the case α ≠ 1 (the non‑Cauchy stable case).","type":"model","stated":"explicit","quote":"Once again we will exclusively look at the α̸ = 1 case.","support":"full"},{"assumption":"For simplicity the two business lines sell an equal number of policies.","type":"data","stated":"explicit","quote":"Further suppose each line sells identical policies whose losses can be (at least asymptotically) 7 described as Pareto-tailed random variables L(1) and L(2) with tail parameters α1 and α2 respectively3.","support":"full"},{"assumption":"A Stable approximation is viable when all business units share the same tail index.","type":"model","stated":"explicit","quote":"Notably, we make use of the fact that as long as our business units share a tail index we can make a Stable approximation.","support":"full"},{"assumption":"Only when the tail parameters of the two lines are equal (α1 = α2) can the aggregate loss vector be modeled as a non‑degenerate Stable vector.","type":"model","stated":"explicit","quote":"As we will see, however, the only case in which we can do so without introducing degenerate marginals is when α1 = α2.","support":"full"},{"assumption":"α > 1 is assumed so that the scaling vector qn can be defined.","type":"model","stated":"explicit","quote":"Since α > 1 we can set qn = (n1−1 nα )(µ1, µ2)T where µi are the respective means.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Data compiled by Eaton et al., 1971 and Embrechts et al., 2013","provider":null,"kind":"other","access":"unknown","period":null,"universe":null,"quote":"Considering the data complied by [Eaton et al., 1971] and [Embrechts et al., 2013] we believe this makes for a more realistic application to insurance.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"none"},"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":"none","findings":[{"finding":"The Tail Conditional Expectation of aggregate risks can be computed within the multivariate Stable framework and allocations derived accordingly.","kind":"theoretical","quote":"We additionally compute the Tail Conditional Expectation of aggregate risks within the model and the corresponding allocations.","support":"full"},{"finding":"Allocations are linear in the aggregate risk capital when the underlying losses follow a bivariate Normal distribution, illustrating a special case of the general approach.","kind":"theoretical","quote":"In such a case allocations were shown to be linear in the aggregate risk capital: A[X(i), S] = E[X(i)] + γX(i),S (H(S) −E[S]) where γX(i),S is the typical regression coefficient, namely Cov[X(i), S]/Var[S] where Cov[X, Y ] and Var[X] are the covariance and variance respectively.","support":"full"},{"finding":"A Stable approximation of the aggregate loss vector is possible only when the two lines share the same tail index (α1 = α2).","kind":"theoretical","quote":"As we will see, however, the only case in which we can do so without introducing degenerate marginals is when α1 = α2.","support":"full"},{"finding":"As long as business units share a tail index, a Stable approximation can be used for allocation with systemic shocks.","kind":"theoretical","quote":"Notably, we make use of the fact that as long as our business units share a tail index we can make a Stable approximation.","support":"full"},{"finding":"The proposed methods provide a very general, prudent and simple way of handling the allocation problem, assuming the worst‑case tail index for each loss.","kind":"theoretical","quote":"The methods presented here allow for a very general, prudent and simple way of handling the allocation problem. By assuming the worst-case tail index for each loss, the Stable approximation is viable.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":15,"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":2,"extracted_at":"2026-11-14T19:54:31Z"},"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/"}