The analysis in End Home Equity Theft measured the impact of property tax foreclosures in states where homeowners cannot get the surplus value (their equity) back after foreclosure.
We measured three issues: (1) the number of homes that were tax foreclosed and sold in these states, (2) the equity the homeowner lost above the tax debt they owed, and (3) the surplus over the tax debt that the government or a private investor kept after the sale. Below, we discuss the data and methodology for the analysis.
Our analysis focused on homes that went through a tax foreclosure sale between January 1, 2014, and December 31, 2021. We did not count homes that were foreclosed but not sold in this timeframe. We only counted residential properties and did not count commercial, industrial, vacant, or other types of property.
This report is based on public records from the most populous taxing jurisdictions in states that allow home equity theft. Our goal was to cover at least 33% of the population in states with city-based taxing jurisdictions and at least 50% of the population in states with county-based taxing jurisdictions (see Table A.1). We also included the taxing jurisdiction that encompasses the capital city.
The inconsistency of public records limited our ability to create a comprehensive dataset, and the data we did collect cannot be extrapolated to the whole state. Public records are often stored in different formats from one jurisdiction to the next, making data extraction and aggregation difficult. For example, some of the jurisdictions that we requested information from only kept paper files of the records we needed. In most cases, these records were copied and sent to us. In a few cases, the records were handwritten in physical books, and the process of obtaining them was too time intensive and costly (see documentation in the state data .zip files).
In all, we have data for 203 taxing jurisdictions across 12 states and Washington, DC.
If we were unsuccessful in acquiring the appropriate records from a taxing jurisdiction, we collected information from the government websites, where available. If we remained unsuccessful, we attempted to find tax-foreclosed properties from the deed transactions collected by Estated, a property data company.1 For example, for some taxing jurisdictions, we were able to identify deeds in the Estated data with the locality as the seller and confirmed that the deed transfer was a tax foreclosure using online property records or a follow-up public records request about those specific properties. Note that Estated has been purchased by Attom Data Solutions. We received data for DC and South Dakota from Attom, however the data structure and collection were still modeled the way Estated had the information.
The report timeframe was from January 2014 through December 2021. However, the specific record dates varied. Data were not available for some taxing jurisdictions for the complete timeframe. Six taxing jurisdictions did not have records going back to January 2014. In addition, for taxing jurisdictions where we started receiving data in late 2020, we have not gone back for updated records. For specific record dates, see the documentation in the state data .zip files.
We supplemented our data collection with property records that we purchased from Estated. Estated collects records from assessors, treasurers, and recorders. We used these records to determine if a property was a home, its latitude and longitude coordinates, and Estated’s estimated value of the home. Arizona was our first state and only had Estated data going through 2019. For Massachusetts, we had Estated data going through April 2021. For Colorado, Minnesota, Nebraska, New Jersey, and Oregon, we had Estated data going through September 2021. The remaining states had Estated data through the timeframe.
In some taxing jurisdictions, the value we used for taxes owed (total due) may not fully reflect the amount that the homeowner would have needed to pay to redeem their property. We used the best available data (see the documentation in the state data .zip files). If we had enough data to do so, we used a multiplier to estimate the total due at foreclosure (see the documentation in the Massachusetts .zip file).
Sale Price and Sale Date
Sale price and sale date primarily came from the locality’s records. When county records were missing or incomplete, we used Estated data for sale price and sale date.
Some homes were sold together in one transaction, which we determined from either public records or the Estated data. When homes were sold together, we divided the sale price by the number of homes in the deed to roughly apportion the total into individual sales prices. This procedure eliminated double counting of sales prices without affecting average calculations.
Estimated Home Value
Home values came from Estated’s proprietary valuation algorithm.2 These values were estimated for the month we received the data. We adjusted all values for inflation to approximate the post-foreclosure sale values. To adjust for inflation, we used the Federal Housing Finance Agency’s House Price Index, Seasonally Adjusted Price Index (HPI).3 We used the formula:
where HPIS is the HPI for the quarter and year when the property was sold, HPIE is the HPI for the quarter and year when the data were obtained from Estated, and V is the property value obtained from Estated.
We examined the number of homes taken for tax debts. We then looked at how much equity homeowners lost (homeowner estimated equity lost) and how much governments kept above what was owed (homeowner equity kept by government/investors).
Homeowner Equity Lost
To estimate homeowners’ lost equity, we used the home’s estimated value (using Estated property data valuations adjusted for inflation). We subtracted the total taxes, interest, and fees due on the home (from public records). Of the 8,589 homes foreclosed and sold in our dataset, we obtained complete records of homeowner equity lost for 6,455 homes.
Homeowner Equity Kept by Government/Investors
To calculate the estimated surplus kept by government and private investors, we identified the price the government or investors sold the home for (using public records or Estated deed transactions) and subtracted the total taxes, interest, and fees due on the home (from public records). To the extent that public records do not include additional costs, our estimates undercounted them (in Massachusetts, we attempted to calculate these costs). Of the 8,589 homes foreclosed and sold in our dataset, we have complete records of homeowner equity kept by government/investors for 5,961 homes.
On the Communities page, we focus on how often home equity theft occurs in an area. To measure the rate of home equity theft, we counted the number of instances of home equity theft within a Census tract over our seven years of data. We took instances of home equity theft and divided them by the total number of housing units in the tract in 2018 (the middle of the period). 1 Communities hardest hit by home equity theft had one or more homes out of every 100 homes foreclosed.
The instances of home equity theft within a tract were counted using the “Count Points in Polygon” algorithm in Quantum Geographic Information System (QGIS) based on the polygons from the Census Bureau’s 2018 Topologically Integrated Geographic Encoding and Referencing (TIGER) program tract shapefiles. 2
However, we undercount the rate of home equity theft in each area. In addition to the constraints highlighted in the sections above, we were constrained to data points with latitude and longitude. For most of the data, Estated/Attom provides the latitude and longitude. In cases where this information was missing, we attempted to find the points using Geocodio. 3 Consequently, a total of 8,312 thefts are represented on the Communities page.
In addition to examining the rate of home equity theft by tract, we investigated the following information from the American Community Survey (ACS) to obtain a bigger picture of these communities. 4
• Median household income
• Unemployment rate
• Percentage of the population that was black and Hispanic
• Percentage of the population with at least a bachelor’s degree
• Percentage of the population with only a high school diploma