CFPB HMDA Mortgage Disclosure Scraper

Pulls loan-level mortgage disclosure records from the CFPB/FFIEC HMDA Data Browser. Returns pricing (rate, points, fees), underwriting (DTI, LTV, credit score type, automated underwriting results), applicant demographics, and census-tract detail for every originated, denied, or withdrawn mortgage application reported under the Home Mortgage Disclosure Act — filterable by year, state, loan outcome, and filer LEI, covering 2018 through 2025.

---

CFPB HMDA Scraper Features

• Returns all 78 loan-level fields per record — pricing, underwriting, applicant, and tract demographics, not just a summary count
• Joins each record's filer LEI to its institution name, so output is usable without a second lookup
• Filters stack in one run: select multiple years, multiple states, and multiple loan outcomes together
• Covers every published HMDA year, 2018-2025, and all 50 states, DC, and US territories
• No account or API key required — the data is public

---

Who Uses HMDA Mortgage Data?

• Mortgage lenders — benchmark competitor pull-through, pricing, and denial rates by state or MSA
• Fair-lending and CRA consultants — run disparity analysis across race, ethnicity, and sex fields
• Bank M&A teams — size a target institution's origination book before a deal
• Proptech and mortgage-fintech teams — pull market-sizing data for a new product or region
• Community development and housing researchers — track lending patterns by census tract over time

---

How CFPB HMDA Scraper Works

1. Choose the years, states, and loan outcomes you want — every filter accepts more than one value.
2. The actor queries the Data Browser for that selection and returns the full loan tape that matches.
3. Each record's filer LEI is resolved to the reporting institution's name automatically.
4. Records stream into your dataset as structured JSON, capped at your maxItems.

---

Input

{
  "years": ["2024"],
  "states": ["DC"],
  "actionsTaken": ["1"],
  "leis": [],
  "maxItems": 500
}

Field	Type	Default	Description
years	array (required)	["2024"]	HMDA activity year(s) to query. 2018 through 2025 — select more than one to pull multiple years in one run.
states	array	(nationwide)	US state/territory code(s) to filter by. Leave empty to query nationwide — a much larger result set.
actionsTaken	array	(all outcomes)	Loan outcome code(s): 1 Originated, 2 Approved not accepted, 3 Denied, 4 Withdrawn, 5 Closed for incompleteness, 6 Purchased loan, 7 Preapproval denied, 8 Preapproval approved not accepted. Leave empty for every outcome.
leis	array	(all filers)	Legal Entity Identifier(s) of specific reporting institutions. Leave empty to include every filer matching the other filters.
maxItems	integer	500	Maximum number of loan records to return.

Filtering to a single state's originated loans for one institution:

{
  "years": ["2023", "2024"],
  "states": ["CA"],
  "actionsTaken": ["1"],
  "leis": ["549300AG64NHILB7ZP05"],
  "maxItems": 1000
}

---

CFPB HMDA Scraper Output Fields

{
  "lei": "549300AG64NHILB7ZP05",
  "institution_name": "LOANDEPOT.COM, LLC",
  "activity_year": 2023,
  "state_code": "DC",
  "county_code": "11001",
  "census_tract": "11001001901",
  "derived_msa_md": "47894",
  "action_taken": 1,
  "loan_type": 2,
  "loan_purpose": 1,
  "loan_amount": 465000,
  "interest_rate": 6.625,
  "rate_spread": -0.1,
  "loan_to_value_ratio": 80,
  "income": 185,
  "debt_to_income_ratio": "36%-<40%",
  "derived_race": "Black or African American",
  "derived_sex": "Joint",
  "applicant_race_1_5": [3],
  "aus_1_5": [1],
  "tract_minority_population_percent": 48.2,
  "scrapedAt": "2026-10-15T07:33:31.931Z"
}

Field	Type	Description
lei	string	Legal Entity Identifier of the reporting institution
institution_name	string	Reporting institution's name
activity_year	integer	HMDA reporting year
state_code / county_code / census_tract	string	Property location (state, county FIPS, tract FIPS)
derived_msa_md	string	Metropolitan statistical area / metro division code
action_taken	integer	Loan outcome code (see the Input table)
loan_type, loan_purpose, lien_status	integer	Conventional/FHA/VA/RHS, purchase/refi/other, first/subordinate lien
loan_amount, property_value, income	number	Dollar figures (income in thousands)
interest_rate, rate_spread, loan_to_value_ratio	number	Pricing figures
total_loan_costs, origination_charges, discount_points, lender_credits	number	Closing-cost breakdown
debt_to_income_ratio	string	DTI bucket or exact value
applicant_credit_score_type, co_applicant_credit_score_type	integer	Credit scoring model used
derived_race, derived_ethnicity, derived_sex	string	Single-value derived demographic categories
applicant_race_1_5, applicant_ethnicity_1_5, co_applicant_race_1_5, co_applicant_ethnicity_1_5	array	Up to 5 reported race/ethnicity codes per applicant
aus_1_5	array	Automated underwriting system(s) used and result code(s)
denial_reason_1_4	array	Denial reason code(s), populated when action_taken is 3
tract_population, tract_minority_population_percent, ffiec_msa_md_median_family_income	number/integer	Census-tract demographics joined in by the source
scrapedAt	string	ISO-8601 timestamp when the record was emitted

78 fields are returned per record in total — the table above covers the ones most buyers filter or join on.

---

FAQ

How do I scrape HMDA mortgage data?

Give the CFPB HMDA Mortgage Disclosure Scraper a year (or several), optionally a state and a loan outcome, and run it. It returns every matching loan-level record as structured JSON — no manual CSV export from the Data Browser required.

What data can I get from the HMDA Data Browser?

Every field the CFPB publishes under Regulation C: loan pricing and terms, underwriting detail (DTI, LTV, AUS results, denial reasons), applicant demographics, and census-tract statistics, for any combination of year, state, and loan outcome from 2018 through 2025.

Can I filter by lender?

Yes. Pass one or more Legal Entity Identifiers in leis to narrow the result to specific reporting institutions. The actor also resolves each LEI to the filer's name automatically.

Do I need an account or API key for the HMDA Data Browser?

No. HMDA data is published by the CFPB as public record, and this actor needs no credentials to query it.

How much does the CFPB HMDA Scraper cost to run?

Cost scales with the number of records returned — narrow your year/state/outcome selection to control it, or set maxItems to cap a run outright.

---

Need More Features?

Need a different filter, a county/MSA-level breakout, or a cross-year rollup? File an issue or get in touch.

Why Use This Scraper?

• Loan-level, not just counts — returns the full 78-field loan tape, not a state/year aggregate.
• LEI-keyed output — every record carries both the filer's LEI and its resolved institution name, so it joins straight into any other LEI-keyed dataset without a manual lookup step.
• One run, many filters — years, states, and outcomes all stack in a single request instead of forcing one run per combination.