{"paper_id":"2411.00563","title":"Simulate and Optimise: A two-layer mortgage simulator for designing novel mortgage assistance products","authors":["Leo Ardon","Benjamin Patrick Evans","Deepeka Garg","Annapoorani Lakshmi Narayanan","Makada Henry-Nickie","Sumitra Ganesh"],"paper_date":"2024-11-01","paper_type":"software-tool","paper_url":"https://arxiv.org/pdf/2411.00563v1","page_url":"https://thequant.space/flowcharts/simulate-and-optimise-a-two-layer-mortgage-simulator-for-de/","scores":{"empirical_rigor":7.5,"math_complexity":6,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a novel two-layer approach that combines a simulation layer of adaptive households with an optimisation layer that designs mortgage assistance products.","quote":"We develop a novel two-layer approach for optimising mortgage relief products through a simulated multi-agent mortgage environment. We develop a two-layer approach for mortgage product optimisa-tion, illustrated in Fig. 1.","summary":"We develop a novel two-layer approach for optimising mortgage relief products through a simulated multi-agent mortgage environment. While the approach is generic, here the environment is calibrated to the US mortgage market based on publicly available census data and regulatory guidelines. Through…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The environment is calibrated to the US mortgage market using publicly available census data and regulatory guidelines.","type":"data","stated":"explicit","quote":"While the approach is generic, here the environment is calibrated to the US mortgage market based on publicly available census data and regulatory guidelines.","support":"full"},{"assumption":"Households are adaptive agents that learn to make mortgage‑related decisions to maximize their utility.","type":"model","stated":"explicit","quote":"Households in the simulation are adaptive, learning to make mortgage-related decisions (such as product enrolment or strategic foreclosures) that maximize their utility, balancing their available liquidity and equity.","support":"full"},{"assumption":"Product parameters (up‑front premium, monthly fee, cover value) are non‑negative and bounded (e.g., fees never exceed the base mortgage payment).","type":"execution","stated":"explicit","quote":"The product is represented as: 𝜑(𝑃0, 𝑃𝑡,𝑉, . . . ) (4) covering three key parameters: the upfront premium 𝑃0 ≥0 (e.g. enrolment cost), the ongoing monthly fee 𝑃𝑡≥0, and the total cover value provided by the product 𝑉≥0.","support":"full"},{"assumption":"Cover provided by a product is not replenished during the simulation episode.","type":"model","stated":"explicit","quote":"0 ≤𝑣≤𝑇adjusts the value of the cover, upper bound to be full cover for the duration of the simulation episode (we assume no cover replenishment here).","support":"full"},{"assumption":"The fixed product layer samples product parameters from a multivariate uniform distribution for simplicity.","type":"execution","stated":"explicit","quote":"For this work, the fixed layer samples from a multivariate uniform distribution 𝐷= {U(0, 1), U(0, 1), U(0,𝑇)} for simplicity, but any distribution could be used.","support":"full"},{"assumption":"The adaptive product layer updates the bounds of the parameter distribution using a reinforcement‑learning based reward signal.","type":"model","stated":"explicit","quote":"While the representation allows for any optimization algorithm to output an updated 𝜽, here, we use a RL-based approach, taking actions a = {Δ𝜇𝑝0, Δ𝛿𝑝0, Δ𝜇𝑝𝑡, Δ𝛿𝑝𝑡, Δ𝜇𝑉, Δ𝛿𝑉} that adjust the boundaries of the underlying multi-variate distribu-tion, where the lower bound is 𝜇−𝛿, and upper bound is 𝜇+𝛿, using the optimization metric −L as reward signal driving the update.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Census and publicly available datasets for household financial profiles","provider":null,"kind":"macro","access":"public","period":null,"universe":null,"quote":"Each borrower has a realistic financial profile including their income 𝐼𝑡, expenses 𝐸𝑡(including mortgage expense $𝑚), and amount of savings 𝐴𝑡, all sampled according to census and publicly available datasets (detailed in Table 1) at 𝑡= 0, as well as a random mortgage term remaining (from 0 . . .","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"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 two‑layer simulation approach can design novel mortgage assistance products that improve household resilience to income shocks.","kind":"other","quote":"We show how this novel two-layer simulation approach can successfully design novel mortgage assistance products to improve household resilience to exogenous shocks, and balance the costs of providing such products through post-hoc analysis.","support":"full"},{"finding":"Optimising for a social index reduces delinquency rates, especially for the most affected borrower groups, thereby increasing financial stability.","kind":"performance","quote":"5 CONCLUSION AND DISCUSSION We developed a novel two-layer approach for automated finan-cial product design, and demonstrated the use of this approach for designing mortgage products in a simulated mortgage ecosystem. Through optimising for a social index, we showed how we can improve the financial stability of households (by reducing delin-quencies), which can lead to improved homeownership rates for even the most","support":"full"},{"finding":"The adaptive product layer successfully identifies parameter regions that improve the social index compared to baseline products.","kind":"performance","quote":"To demonstrate that the adaptive product layer works as intended (finding a suitable 𝜽), we evaluate across the social index, showing that a logical region is found that optimises L. L(𝜔(𝑟)).","support":"full"},{"finding":"The integrated metrics show that the outer layer can find products that improve both profit for providers and social welfare metrics.","kind":"performance","quote":"In all cases darker colour indicates a better value. in the simulation, instead a balance is preferred, and providers can still see good profit from higher cover products, demonstrating one benefit of the simulation for modelling product uptake. The inte-grated metrics are presented in Table 3, demonstrating the outer layer successfully finds products which improve over the baseline under both metrics.","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":12,"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":"2026-11-14T19:54:21Z"},"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/"}