Check the setup
- Python 3.10 or newer
- A notebook environment such as Jupyter or Google Colab
- An internet connection
Home Mortgage Disclosure Act loan-application records for building local mortgage-market and fair-lending research tools.
From source to product signal
HMDA LAR records loan applications, originations, and denials. Start with one year and one state through the Data Browser. Applicant race and ethnicity fields have disclosure rules, and a LAR row is not a credit decision you can reverse-engineer.
Install the packages, then run the notebook cell.
python -m pip install pandas requests
from io import StringIO
import pandas as pd
import requests
response = requests.get(
"https://ffiec.cfpb.gov/v2/data-browser-api/view/csv",
params={
"years": "2023",
"states": "VT",
"actions_taken": "1",
},
timeout=120,
)
response.raise_for_status()
loans = pd.read_csv(StringIO(response.text), nrows=200)
print(loans.head())Test a useful signal
Test whether a one-state LAR extract can power a bounded mortgage-volume sketch.
Consumer Financial Protection Bureau is a government source. Last verified 2026-08-18. Temporal coverage: annual HMDA filings from 2007-present with schema changes.