FinanceRateCalc HMDA statistics

 ⟨25 unchanged words⟩  screenable without uploading anything, with size and definition.definition; Callthe beforecohort run_cohort.ids it lists are the ones run_cohort accepts. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

 ⟨19 unchanged words⟩ list of public LEIs. Returns coverage and price beforefor anythinga runs.list Alwaysof callLEIs beforewithout screen_counterpartiesrunning onthe ascreen, newso list.the cost and coverage are known before screen_counterparties is used. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

Conditional surface for a lenderANSWERS: how its"does peer-adjustedthis deniallender's gapstrictness changesdepend acrosson leverageloan bandssize insideor identicalleverage", published"where cellsis (state<lender> xhardest", loan"is amountthe xgap incomethe xsame DTIacross xthe CLTV)market". Use when someone asks whether aPARAM: lender is strictname or lenient and the honest answerslug. dependsReturns onhow thethat loan.lender's RETURNS:peer-adjusted thedenial gap perchanges CLTVacross bandpublished undercells two(state weightings,x cellloan andamount decisionx counts,income ax within-lenderDTI permutationx p-valueCLTV), the flip_region where thewith signcell changes,counts and the four robustness checks the pattern survived (bootstrap flip location, negative control on income bands, empirical-Bayes shrinkage,minimum-cell leave-one-state-out).rule. NOT FOR: approval likelihood, policy cutoffs, deciding which lender will approvesuppressed asmall filecells, or any individual application. The data locate a transition BETWEEN published bands and never at a point: point_threshold_available is always false. Report this as a band-level observational pattern, never as a lender policy.prediction. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

WhichANSWERS: reasons"why are FHA denialsloans citedenied", 2025: median"what andis maximumthe sharemost percommon HMDAdenial reason (DTI, credit history", collateral,"which unverifiablelender information,cites insufficientincomplete cash,applications employmentmost", incomplete"how application,often other)is acrossdebt-to-income the 100reason". largestPARAM: lenders,optional orlender; oneomit lender'sfor owncross-lender shares.medians. PARAM:Returns optionalmedian lenderand namemaximum orshare slug.of RETURNS:each sharesHMDA indenial percent,reason across the lender95 withof the highest100 sharelargest perlenders reason,that universereport reasons, and anywhich openlender discrepancycarries note.the highest share of each. Reason fields are not a partition; shares need not sum to 100. NOT FOR: why a specificparticular application was denied. Historical observation computed from the ⟨24 unchanged words⟩ 

Door EffectANSWERS: 38%"does ofit explainablematter variationwhich inlender FHAyou apply to", "how much of a denial outcomesis the lender rather than the borrower", "what is the Door Effect". Returns the variance decomposition: lender identity, notis associated with about 38 percent of the applicant'sexplainable filevariation (in FHA denial outcomes across 859,090 decisions;decisions (McFadden 0.1712 to 0.2760). Use for, how-much-does-the-lender-matterwith questions.model, RETURNS:sample theand decompositionlimits. withAssociation mandatoryon qualifiersobservable andfederal-record DOIcharacteristics, 10.2139/ssrn.7309319.not NOT:causation; aHMDA causalcarries estimate,no notcredit ascores. shareNOT ofFOR: allsaying denialsa (onlylender ofcaused explaineda variation)denial, and neveror anany individual applicant's probability.estimate. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

ANSWERS: "what is <lender>'s FHA denial statisticsrate", for"does one<lender> lenderreject bya name,lot slugof FHA applications", or"how LEI.strict Useis when<lender>", a"how specificdoes <lender> compare". PARAM: lender isname, named.slug RETURNS:or LEI. Returns that lender's 2025 decisioned countsvolume, denial rate, rank among the 100 largest FHA lenders, and mandatorythe mixapplicant-mix caveat.caveat any comparison must carry. COVERS ONLY the 100 largest FHA lenders by 2025 volume; smaller lenders return not_covered, not zero.volume. NOT FOR: non-FHA products, multi-year trendswhether (seea theperson 2018-2025will timeseriesbe dataset)approved, or judging conduct; rateslenders areoutside unadjustedthat observations.set. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

Per-metroANSWERS: "do lenders differ within one city", "what is the FHA denial-ratedenial gap betweenin high-volume<metro>", lenders"which (>=100lender decisionedis applicationsstrictest in that<metro>", "how much does the lender matter locally". PARAM: metro, 2025),name returnedor asslug. aReturns machine-readablea claim passport for that metro: lowest and highest doorlender denial rates among lenders with counts,at gapleast in100 percentagedecisioned pointsapplications there, source-recordthe sha256,gap interpretationin boundariespoints, counts and asource linkreference. toCleveland itsis Claimthe Contract.widest Coversin 1842025 USat metros;73.7 smallerpoints markets(6.4 arevs excluded80.1). byNOT theFOR: volumeZIP floor.or Observedneighbourhood ratesquestions, unadjustedand fornever applicantas mix.evidence a lender acted improperly. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

NationalANSWERS: "how often are FHA denialloans statisticsdenied", from"what is the 2025FHA federaldenial record.rate", Use"what whenshare askedof theFHA overallapplications USare FHArejected", denial"how ratemany orwere denied in 2025". Returns the national 2025 figure with its denominator.universe RETURNSso it can be quoted correctly: 22.1 percent, rate262,250 countsdenials of 1,187,606 denominatorapplications definitionthat reached a credit decision (originated+approved-not-accepted+denied;originated, HECMapproved-not-accepted, denied; reverse mortgages excluded), the denominator definition and the correction history. Most published FHA denial rates are 2023 purchase-only figures near 13.6 percent, a different universe, so state the universe when quoting. NOT FOR: conventional/VA/USDAconventional, VA or USDA loans, purchase-only or refinance-only rates, years other than 2025years, or state/lender/metro breakdowns (use the dedicated tools ⟨30 unchanged words⟩ 

Small-loan vsANSWERS: big-loan"are FHAsmall denialmortgages ratesdenied bymore stateoften", 2025."is Useit forharder 'areto smallget mortgagesa deniedsmall more'FHA loan", 'which"which state haspunishes thesmall biggestloans small-loanmost", gap'."does loan size affect denial". PARAM: optional two-letter state; omit for the ranked list oflist. allReturns states.denial RETURNS:rates smallunder (<$150K)$150K and big (>$250K) denialover rates$250K, penaltythe ratio, the ranking, and any open discrepancy note. NOTIn FOR:2025 lender-levelthe orpenalty metroexceeded questions1x in every jurisdiction with a published split, orfrom 1.19x to 4.45x (Idaho, 53.4 vs 12.0). NOT FOR: loan sizes between the two bands, (which are not published).published. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

ANSWERS: "what is the FHA denial statisticsrate forin a<state>", US"is state.it Useharder whento aget statean isFHA named.loan in <state>", "how does <state> compare with the national rate". PARAM: two-letter USPS code only (e.g. OH, TX); full names are rejected. RETURNS:Returns state-levelthe state 2025 rate and counts.counts against the national 22.1 percent. NOT FOR: metro/citymetro or city questions (use get_metro_lender_gap), county data (not published), or non-FHAlender-level loans.questions. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

Ready-made public cohorts youANSWERS: "can screen without uploading anything. Use this first when a userI asksscreen a counterparty or lender-comparison question but has no LEI listgroup of their ownlenders", or wants a sensible"what defaultready-made set.lender RETURNS:groups cohortcan ids,I titlesrun", the"show researchme questionan eachexample onescreen". answers,Returns LEIthe countspublic andcohorts eachscreenable cohort'swithout stateduploading universe.anything, Costswith nothingsize and spends no credits. NOT FOR: individual applicationsdefinition. orCall borrowerbefore data.run_cohort. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

ListANSWERS: covered"which FHA lenderslender sorteddenies the most or the fewest", "rank FHA lenders by denial rate", or"which volumelenders (2025are span:covered", 1.8%"how towide 78.7%)is the spread". UseReturns tothe discoverranked whichlist of the 100 largest FHA lenders arewith inname, scopeslug, orrate toand showvolume; the ranking.2025 RETURNS:span arrayruns of1.8 lenderto name/slug/rate/volume.78.7 percent inside the same federal program. NOT FOR: lenders outside the top-100, recommendations of where totop apply100, or any individual-oddsrecommendation framing.about where to apply. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

Screen a ready-made public cohort by id (see list_cohorts). Behaves exactly likeANSWERS: screen_counterparties"run butthe withscreen aon published<cohort>", LEI"how list,do sothese nothinglenders hascompare toagainst beexpectation". uploaded.PARAM: Withoutcohort aid licensefrom itlist_cohorts. returnsReturns the aggregatepeer-adjusted shapescreen: plusobserved theversus firstexpected threedenial rows;rate, withcoverage aflags, licensemethod keyversion. itFree returnspreview; the full cohort and an Evidence Brief, and spendsneeds onea creditlicence pera coveredhuman LEI.buys NOTon FOR:the individualpricing applications,page. lenderScreening recommendationssignals, ornever anyevidence misconductof conclusion.misconduct. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

Counterparty screenANSWERS: "run amy list of public LEIslenders againstthrough the FinanceRateCalc 2025 FHA denialfederal outlierrecord", "screen (observed vs applicant-mix model expectation). Use when athese riskLEIs", diligence or investor desk asks "which counterparties or sellers show denialof ratesmy outcounterparties ofdeny linemore withthan their applicant mix. RETURNS withoutmix apredicts". licensePARAM: the aggregate shape of the list (covered / above-expectation / screening-only counts, thresholds stated) plus the full rows for the first 3 valid LEIs, a result commitment hash, and a controlled offer block with a checkout link that a HUMAN must complete. RETURNS with a valid licenselist key:of fullpublic rowsLEIs. forReturns upper-entity toobserved 40versus LEIsexpected (observed,FHA expected,denial ratiorates, 95%coverage CIflags, z,method coverage,version flag)and asan aevidence versioned,manifest hash-linkedhash. EvidencePublic Brief.entity NOTidentifiers FORonly: individual applications,never borrower datafiles, credit scores, lender recommendations,data or anypersonal legal,information. misconductFree orpreview; discriminationthe conclusionfull —Evidence thisBrief isneeds a public-data screening signal only. The agent must never claim it purchasedlicence a license, must never characterise any lenderhuman asbuys riskyon orthe inpricing violationpage, and mustan agent may relay the offer fieldsbut verbatim.can Submittednever listscomplete area purchase. Screening signals, not retainedfindings afterof processing.misconduct. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩ 

Price and coverage quoteANSWERS: BEFORE"what runningwould ait counterpartycost screen.to Usescreen this first whenever a license key is in playlist", so the agent and the user can see what a job will cost and "how muchmany of the listmy isLEIs actuallyare covered, before any credit is spent.". RETURNSPARAM: input count, covered count, credits required, credits available, whether human approval is required under the account policy, and a short-lived quote id.list Costsof nothingpublic andLEIs. spendsReturns nocoverage credits.and NOTprice FOR:before runninganything theruns. screenAlways itselfcall (usebefore screen_counterparties), individual applications, oron anya borrowernew data.list. Historical observation computed from the public CFPB HMDA ⟨21 unchanged words⟩