Methodology

The basis of the Home Mortgage Explorer is Home Mortgage Disclosure Act (HMDA) data. Details on HMDA data can be found in the Data Source section. In addition to the HMDA data, variables from the American Community Survey and decennial census are added in the creation of the application rate loan measure and the two loan characteristics: applicant income and neighborhood income. The method of how the loan measure and loan characteristics are calculated in the Home Mortgage Explorer may differ from other published estimates; therefore, this methodology outlines the specific steps taken by the authors of this tool to get the final outputs.

Notes on Connecticut Geography Changes

Connecticut transitioned from counties to planning regions starting in the year 2023. This geographical change first appears in the HMDA data in 2024. Owing to there being only one year of data in the newly classified planning regions, they are not included in this version of the tool. Instead, the data in the planning regions were mapped into the original county definitions using census tract information.¹ In the future, when there are sufficient data using the planning region definitions, they will be included separately in the county geography level.

Owner-Occupied Units for Application Rate

The application rate is the first loan measure in the tool; this variable gauges the rate of applications per 1–4-family, owner-occupied housing unit. This is calculated as the total number of applications as the numerator and total 1–4-family, owner-occupied units in the reference geography as the denominator. The rate is calculated for each year, specified by geography and loan type, multiplied by 1,000. The numerator is derived directly from the HMDA data. The denominator is from the American Community Survey (five-year estimates), except for the 2010–2011 HMDA data, which use the 2000 decennial census. The five-year estimates are used because they provide an accurate owner-occupied unit count at the county level. As the HMDA data are reported yearly, we have aligned the following census estimates to groups of singular HMDA years, corresponding to the census data products that are used by the Federal Financial Institutions Examination Council (FFIEC) for each year of HMDA data:

• 2010–2011 HMDA data use the 2000 decennial census
• 2012–2016 HMDA data use the 2006–2010 American Community Survey
• 2017–2021 HMDA data use the 2011–2015 American Community Survey
• 2022–2024 HMDA data use the 2016–2020 American Community Survey

The corresponding years are merged into the HMDA data at the county level, then aggregated up to the metro or state level. It is important to note that the aggregation at the metro level is aligned with the metro HMDA that categorized that specific county for that year. This decision largely corresponds to the FFIEC publication of metro boundaries.

Median Family Income for Applicant and Neighborhood Income

Median family income is used to determine the income classification of an applicant or neighborhood. Applicant income is reported directly from the HMDA data, while neighborhood income is the median family income of the census tract directly from the American Community Survey (five-year estimates) or the 2000 decennial census. Neighborhoods are classified as the census tract that the loan is associated with. For example, if the application is a home purchase loan, the neighborhood would be the census tract in which the home is located.

Using income information, two classifications are created: low- to moderate-income (LMI) or middle- to upper-income (MUI). LMI is classified as an applicant income or census tract income below 80 percent of the metro’s median family income. MUI is at or above 80 percent of the metro’s median family income. The FFIEC’s estimation of the median income at the metro level is used for the ratio to determine LMI versus MUI status. The FFIEC’s definition is generally aligned with the U.S Department of Housing and Urban Development (HUD) methodology. The difference between the two is when metro boundary changes occur. Then the FFIEC will have its own calculation for the median family income. For more information about the specifics of the median family income calculation, see www.ffiec.gov/data/census/data-dictionary.

Studying mortgage lending trends by income over time can be a challenge because the income used to determine LMI-MUI status could change based on changes in metro boundaries. Metro boundaries change when counties are added or removed based on changes in the population or commuting patterns. As a result, when metro boundaries change, an applicant in one year may be classified differently than a similar applicant in a subsequent year because of the boundary change’s effect on the area metro’s median family income (MFI). Owing to boundary changes for some metros, the data trends will not be comparable over time. The authors created an Excel spreadsheet to document the changes to metros across time for users to better interpret the charts in the tool.