{"paper_id":"2508.10273","title":"A 4% withdrawal rate for American retirement spending, derived from a discrete-time model of stochastic returns on assets and their sample moments","authors":["Drew M. Thomas"],"paper_date":"2025-08-14","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2508.10273v2","page_url":"https://thequant.space/flowcharts/a-4-withdrawal-rate-for-american-retirement-spending-deriv/","scores":{"empirical_rigor":6,"math_complexity":7.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a discrete‑time model of retirement portfolio returns that yields a closed‑form optimal withdrawal rate based on an s‑adjusted rate of return γ.","quote":"I address that question with a discrete-time model of returns to a retirement portfolio consumed at a rate that grows by $s$ per period.","summary":"What grounds the rule of thumb that a(n American) retiree can safely withdraw 4% of their initial retirement wealth in their first year of retirement, then increase that rate of consumption with inflation? I address that question with a discrete-time model of returns to a retirement portfolio…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Returns are a stationary process with constant moments across periods.","type":"statistical","stated":"explicit","quote":"I normalize the 1 + rj terms by first assuming that r is a stationary process, which entails that r’s moments, including the expected value, are constant in all time periods."},{"assumption":"Returns are independently distributed across periods.","type":"statistical","stated":"explicit","quote":"Assuming next that returns are not only stationary but independently distributed (i.e. that the return to wealth in period j provides no information about the return to wealth in any other period)."},{"assumption":"Leverage ratio l is non‑negative, finite and s > –1.","type":"model","stated":"explicit","quote":"Assuming s > −1, l non-negative and finite, and E[r] non-negative and finite, the first derivative is zero when the numerator of its fraction is zero..."},{"assumption":"ETF price series reflect actual fees and can be used as proxies for asset returns.","type":"data","stated":"explicit","quote":"Actual ETF prices offer the advantage of reflecting the actual fees charged to hold the relevant collections of assets."},{"assumption":"Shiller’s long‑run series provides a reliable historical return series back to 1871.","type":"data","stated":"explicit","quote":"I therefore pay more attention to the much longer, albeit coarser‑grained, time series of Shiller, which extend back through 1871."},{"assumption":"Margin borrowing cost equals the effective federal funds rate plus a 100‑basis‑point spread.","type":"execution","stated":"explicit","quote":"charge 90–100 basis points per year above the annual EFFR as of mid‑2025."}]},"datasets":{"status":"available","items":[{"name":"ETF monthly returns (S&P 500 and US Treasury Bonds)","provider":"Yahoo! Finance","kind":"market-prices","access":"public","period":"spring 2024 (downloaded)","universe":"oldest ETFs representing the S&P 500 index and US Treasury Bonds","quote":"I selected the oldest ETFs I could find that represented the S&P 500 index of US equities, and US Treasury Bonds of different durations.","known_id":"yahoo-finance"},{"name":"Shiller historical return series","provider":"Robert Shiller","kind":"market-prices","access":"public","period":"1871‑present","universe":"US equities and Treasury bonds","quote":"I therefore pay more attention to the much longer, albeit coarser‑grained, time series of Shiller, which extend back through 1871."}],"known":[{"id":"twitter","name":"Twitter / X","provider":"X Corp.","kind":"text-news","access":"public","quote":"X/Twitter post posted 9 November 2023, and retrieved from https://x.com/mbontrager5/ status/1722478848573329702 on 15 November 2025.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"weekly-monthly"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":null,"detail":"Margin borrowing cost modeled as the effective federal funds rate plus a 100‑basis‑point spread (90–100 bps above EFFR as of mid‑2025).","quote":"charge 90–100 basis points per year above the annual EFFR as of mid‑2025.","source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Margin borrowing cost modeled as the effective federal funds rate plus a 100‑basis‑point spread (90–100 bps above EFFR as of mid‑2025).","card_quote":"charge 90–100 basis points per year above the annual EFFR as of mid‑2025.","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The 4% rule emerges when high expected returns are adjusted down for consumption growth, variance, kurtosis, longevity risk, and bond inclusion.","kind":"theoretical","quote":"γ … reveals that the 4% rule emerges from adjusting high expected rates of return down for: consumption growth, the variance in (and kurtosis of) returns to wealth, the longevity risk"},{"finding":"Leverage ratios greater than 1.6 are historically optimal for portfolios dominated by the S&P 500.","kind":"theoretical","quote":"The model supports leverage of retirement portfolios dominated by the S&P 500, with leverage ratios > 1.6 having been historically optimal under the model's approximations."},{"finding":"Using high‑frequency ETF data gives a first‑year withdrawal rate of about 6.4% for a 30‑year retirement, but long‑run Shiller data lower it to about 6.0% (or 5.2% for a 50/50 portfolio).","kind":"performance","quote":"Switching to Shiller’s long‑run data produces a lower γ4 of 0.00356 … and"},{"finding":"Optimal leverage levels depend on borrowing cost; with a 100‑bp spread the optimal l is 3.09 for a 60/40 portfolio, 1.92 for equities, and 0.10 for bonds.","kind":"theoretical","quote":"the optimal ls are 3.09 for the 60/40 portfolio, 1.92 for equities alone, and just 0.10 for bonds alone."},{"finding":"Even with cheap leverage, withdrawal rates above 6.6% per year could be rendered reasonably safe in simulations.","kind":"performance","quote":"could have rendered withdrawal rates of 6.6% per year or more reasonably safe."}],"reported_performance":null,"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":46,"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":"2026-11-14T17:29:54Z"},"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/"}