Git Product home page Git Product logo

hmda.historical's Introduction

Historical Data from the Home Mortgage Disclosure Act (HMDA)

Contents

  1. Introduction
  2. Data Description
  3. Cleaning Methodology
  4. Codebooks
  5. Citation

Introduction

This repository aims to provide historical mortgage lending data pursuant to the Home Mortgage Disclosure Act (HMDA, 12 USC § 29).

These data are provided as-is here in order to further discussion in U.S. housing policy and the economics literature.

Data on American housing are as important as ever in current public policy debates. My goal is to advance these debates by making historical data accessible and wide open for researchers.

If these data are helpful to you or if you notice any errors, please drop me a line. As the great people at IPUMS decry: use it for good!

Data Description

The data provided are a lightly cleaned copy of the HMDA data spanning (for now) 1981-2006. It is important to note that HMDA reporting and data release standards over this time period changed substantially. In addition, the geographies represented and financial institutions reporting changed as well. It is important to take these changes into consideration in any analysis of the HMDA data over time. Given these caveats, my objective is to provide these the HMDA data in an easily accessible, machine-readable format (pipe | delimited).

Table 1. Data Availability by Reporting Year

Reporting Year Tabular LAR Panel Transmittal Sheet Lowest Unit
1981 X Tract
1982 X Tract
1983 X Tract
1984 X Tract
1985 X Tract
1986 X Tract
1987 X Tract
1988 X Tract
1989 X Tract
1990 X X X Borrower
1991 X X X Borrower
1992 X X X Borrower
1993 X X X Borrower
1994 X X X Borrower
1995 X X X Borrower
1996 X X X Borrower
1997 X X X Borrower
1998 X X X Borrower
1999 X X X Borrower
2000 X X X Borrower
2001 X X X Borrower
2002 X X X Borrower
2003 X X X Borrower
2004 X X X Borrower
2005 X X X Borrower
2006 X X X Borrower

Cleaning Methodology

1981-1989 Data

The 1981-1989 HMDA are from the ASCII-converted files published by the National Archives. These data are the least detailed HMDA data available and are only tabulated by reporting year, lending institution (mortgage companies of bank holding companies were added in 1988), and census tract.

These data are provided as-is. I have not yet performed any data validation steps to examine the quality of these data. There are various "validity flags" in the data to identify whether the tract information is valid, within a tolerance, zero, or invalid. Pending a deeper evaluation of these data, I highly recommend consulting these fields when performing any analyses using the older HMDA data.

Further, some of the ASCII characters did not carry over properly when converted by the National Archives. In these cases you will see bank names listed as [non-standard characters ]; however, the banks still have a reported respondent_id. I will perform further analyses to match the respondent_id for each missing bank to its proper name in the future.

For each year I did my best to harmonise the variable names to easily match and append each file. Unfortunately, the 1981-1989 HMDA data structure is substantially different than the future data and the data are tabulated above the borrower level (in contrast to later HMDA releases). In these years I applied a simple nomenclature to the tabular data fields that work with naming requirements in other statistical software.

1990-2006 Data

1990 saw the first restricted release of the HMDA Loan/Application Register (LAR). The LAR contains detail information about loan applications, their details, their ultimate outcomes, the underlying applicants/borrowers, and the underlying properties. Since 1990, HMDA provides an incredible degree of detail on home mortgage loans at very fine levels of geography, namely the census tract. Properties corresponding to each loan application are coded to census tracts. Census tract numbers are assigned by the Census Bureau and financial institutions reporting HMDA data are instructed as to which which decennial census tract numbering scheme to use, i.e. the 1980 Census, 1990 Census, etc. It is important to read the documentation to know which tract numbering scheme is employed in each HMDA reporting year.

The 1990-onwards HMDA data also provide detailed information of the financial instituions that reported HMDA data. These data are provided in two sources: the transmittal sheet (TS) and reporter panel (panel). These information sheets provide information on each HMDA respondent, such as the institution's office location, federal tax ID, and Federal Reserve RSSD_ID. These data help link the HMDA data to additional data sources, such as the Call Reports, Community Reinvestment Act (CRA) disclosure data, Summary of Deposits (SOD) branch location data, Small Business Administration 7(a) (SBA) lending data, etc.

Table 2. Census Tract Delineations by HMDA Reporting Year

Reporting Year Delineations
1981-1993 1980 Census
1994-2003 1990 Census
2004-2006 2000 Census

Codebooks

Please see the documentation for the variable fields and codes here.

Citation

Please cite the data:

Forrester, Andrew C. Historical Home Mortgage Disclosure Act (HMDA) Data. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2021-10-10. https://doi.org/10.3886/E151921V1

hmda.historical's People

Contributors

acforrester avatar

Stargazers

Kyle F Butts avatar

Watchers

James Cloos avatar  avatar

hmda.historical's Issues

Codebooks

Develop codebooks for the 1990-2003 and 2004-2006 data.

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.