{"paper_id":"2609.25965","title":"Modeling interest rate swap volatility with GARCH processes","authors":["Michał Balcerek","Michał Wronka"],"paper_date":"2026-09-22","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2609.25965v1","page_url":"https://thequant.space/flowcharts/modeling-interest-rate-swap-volatility-with-garch-processes/","scores":{"empirical_rigor":8,"math_complexity":7.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes modeling the conditional volatility of the USD 1Y×10Y forward swap rate using GARCH(1,1), GJR‑GARCH(1,1) and a two‑regime Markov‑switching GARCH (MSGARCH) model.","quote":"We examine the conditional volatility dynamics of the USD 1Yx10Y forward swap rate using GARCH(1,1), GJR-GARCH(1,1), and a two-regime Markov-switching GARCH (MSGARCH) model.","summary":"We examine the conditional volatility dynamics of the USD 1Yx10Y forward swap rate using GARCH(1,1), GJR-GARCH(1,1), and a two-regime Markov-switching GARCH (MSGARCH) model. The analysis uses daily data from 2007 to 2023 and incorporates market-implied measures (ATM swaption volatility and the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Conditional heteroskedasticity (ARCH effects) is present in the swap rate log changes.","type":"statistical","stated":"explicit","quote":"Engle’s test (with default one lag) rejects the null hypothesis of no ARCH effect in favour of the alternative."},{"assumption":"The conditional variance can be modeled in discrete time using GARCH‑type specifications.","type":"model","stated":"explicit","quote":"In this article, we deal with GARCH family of models which feature convenient and accurate equations for volatility expectations."},{"assumption":"The intercept ω is allowed to vary over time, reflecting structural changes.","type":"model","stated":"explicit","quote":"Standard GARCH and GJR‑GARCH models show stable short‑run parameters, but the intercept ω varies markedly across rolling windows, causing instability in the implied long‑run variance."},{"assumption":"Innovation distributions may be non‑Gaussian (Student‑t, GED, Johnson SU).","type":"statistical","stated":"explicit","quote":"Four candidate families widely used in the volatility literature are considered: Gaussian, Student‑t, GED"},{"assumption":"Regime‑switching captures volatility shifts better than single‑regime GARCH.","type":"model","stated":"explicit","quote":"The Nyblom joint tests strongly reject constant‑parameter specifications, pointing towards volatility regime shifts that single‑regime GARCH models cannot accommodate."}]},"datasets":{"status":"available","items":[{"name":"Daily fair swap rates","provider":null,"kind":"market-prices","access":"public","period":"2007‑2023","universe":"USD 1Y×10Y forward swap rate","quote":"Section 3 describes the data set, which consists of daily fair swap rates and European swaption implied volatilities used in the analysis."},{"name":"European ATM swaption implied volatilities","provider":null,"kind":"options","access":"public","period":"2007‑2023","universe":"USD 1Y×10Y forward swap rate","quote":"Section 3 describes the data set, which consists of daily fair swap rates and European swaption implied volatilities used in the analysis."}],"known":[{"id":"cboe","name":"Cboe","provider":"Cboe Global Markets","kind":"options","access":"commercial","quote":"(2) We use strikes predefined in the CBOE specifications and read volatilities directly off the surface and similarly apply relevant index formula.","mentions":4,"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":"forecast-accuracy","findings":[{"finding":"MSGARCH delivers the highest log‑likelihood and lowest AIC among the models, while BIC favours the more parsimonious GJR‑GARCH.","kind":"statistical","quote":"It delivers the highest log‑likelihood and lowest AIC, whereas BIC favours the more parsimonious GJR‑GARCH."},{"finding":"The intercept ω exhibits strong time‑variation, whereas short‑run parameters (α, β, γ) remain stable.","kind":"statistical","quote":"Nyblom tests (Section 5.3.1) show that the short‑run parameters (α, β, γ) remain stable, while the intercept parameter ω exhibits strong time variation."},{"finding":"Johnson SU innovations provide the best overall fit among tested distributions, though the null is still rejected at conventional levels.","kind":"statistical","quote":"By contrast, the Johnson SU family provides by far the best overall fit, capturing both skewness and heavy tails, although the null hypothesis is still rejected at conventional levels."},{"finding":"Both MSGARCH and standard GARCH filters identify the same high‑volatility episodes (2008 crisis, 2011‑12 Eurozone stress, COVID‑19 shock) and low‑volatility periods, with MSGARCH sharpening regime transitions.","kind":"performance","quote":"Both approaches identify the same high‑volatility episodes (the 2008 crisis, the 2011–2012 Eurozone stress, the COVID shock) and the same extended low‑volatility period in the mid‑2010s, with MSGARCH mainly sharpening the transitions between regimes."},{"finding":"One‑day‑ahead forecast coverage remains close to the nominal 95% level for standard GARCH and GJR‑GARCH models.","kind":"predictive-accuracy","quote":"Rolling‑window backtesting (Section 7) confirms that one‑day‑ahead forecast coverage remains close to the nominal 95% level, demonstrating that these models perform well for short‑horizon volatility prediction."}],"reported_performance":null,"out_of_sample":{"method":"walk-forward","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":48,"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":1,"extracted_at":"2026-11-14T17:04:45Z"},"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/"}