Towards a National Housing Loss Rate

In October of 2023, New America and Rockefeller Foundation convened a group of leaders in the housing and homelessness, economic development, civil rights, and statistics fields to discuss the potential of developing a National Housing Loss Rate to stand alongside the national unemployment rate as a key metric of social and economic wellbeing. The discussion focused on a survey-based approach for arriving at a National Housing Loss Rate.

This blog post recounts the major discussion points, ideas, and outcomes surfaced during the meeting. Per the Chatham House rules of the convening, all ideas are presented without attribution.

Each year, more than 10 million Americans lose their homes through eviction and foreclosure, tax sales, eminent domain, post-disaster displacement, and other less-studied forms of housing loss. These destabilizing events lead to homelessness, job loss, adverse health impacts, and downward economic mobility. And yet, America neither tracks housing loss (also known as forced residential displacement), nor holds politicians and decision-makers accountable for keeping it low.

It is said you can’t fix something you can’t measure. If America wants to get serious about addressing the disruption, destabilization, and long-term negative impacts caused by housing loss, then it needs to track the number of people who lose their homes each year.

Just as America has a national unemployment rate, it should establish and track a National Housing Loss Rate as a key indicator of social and economic well-being.

A housing loss metric, if rigorous, regularly-collected, and available at the national, state and local level, would have profound impacts on our understanding of the causes and consequences of home loss, and improve our ability to develop policies and programs that keep people more securely housed.

We define housing loss as a measure of residential displacement, or the number of individuals or households that lose their homes involuntarily over a specified period of time.

In other words, an instance of housing loss occurs when a resident loses their home involuntarily, for example through an eviction, foreclosure, or destruction from a wildfire or storm. This distinction is important because the unit of measure is the individual (or household) as opposed to the housing structure.

However, there is nuance within this definition, and the way in which we define housing loss can lead to vastly different measures of the problem. During our convening, we discussed some of the definitional questions that need to be addressed, including:

• What constitutes “loss”? Is temporary displacement (e.g., after a hurricane) considered housing loss? At what point in time does temporary displacement become permanent?
• What constitutes “involuntarily”? Are moves due to economic pressures (such as rent increases and gentrifying forces) considered housing loss? Is forced vs. responsive the most important distinction in defining housing loss?
• Does the presence of compensation, for example in the case of an eminent domain seizure or a managed retreat buy-out, impact whether the move is a “home loss”?
• Is it more important to track the housing loss event, or whether housing is stable?
• Is a singular number (i.e., a housing loss rate) sufficient in establishing an actionable benchmark, or is it necessary to develop a more detailed measure (and if so, what level of detail is adequate, and what are the tradeoffs between feasibility and detail)? What would a measure of housing loss need to tell us, and why?

In small groups, we discussed what a minimum viable measure of housing loss would need to look like, in order to be useful. Small group discussions generally coalesced around the following criteria and identified existing housing-related survey efforts that met each of the criteria below:

• Frequency: A measure of housing loss must be as dynamic as possible to accurately capture how housing loss unfolds. A survey administered quarterly is ideal, although an annual rate could work. Only the Household Pulse Survey (biweekly) and the Robin Hood survey in NYC (quarterly) are conducted on a basis more frequent than annually.
• Geographic Scale: A Housing Loss Rate would ideally be produced at the national level as a benchmark indicator that can inform federal housing policy and drive nationwide momentum. At the same time, most housing decisions and policies are developed at the state and local level, and more granular data will be necessary to inform housing decisions. Of relevant federal surveys, the Pulse Survey collects data at the state level, and the American Community Survey (ACS) provides geographic specificity (down to the neighborhood level) nationwide.
• Scope: Data collected would ideally cover not only the loss but also the cause of the loss. Questions arose as to whether capturing housing loss as one of several forms of housing insecurity – alongside lack of affordability, lack of safety, and poor housing quality – makes sense. HUD PD&R’s Housing Security Index, the Household Pulse Survey, and the Robin Hood Poverty Tracker survey on forced moves in NYC all cover some part of this scope.
• Target Population: A survey needs to reach respondents experiencing housing loss (or who have recently experienced housing loss). This is challenging because transient populations are difficult to reach through household surveys, and also because most existing federal surveys target a non-institutionalized population and exclude people in group quarters who may have higher rates of housing insecurity (e.g., people who are incarcerated or staying in homeless shelters and group homes). More generally, housing loss can be hard to assess because surveyors often lack access to contact information for people who have moved.

In an ideal case, a survey that measures housing loss should be:

• Administered at the household level;
• Longitudinal;
• Administered frequently;
• National in scope, with estimates produced at the state and local levels; and
• Suited to sample a hard-to-reach population.

We reviewed several existing survey instruments that meet some of the above criteria. No survey efforts currently meet all of the criteria.

The group spent significant time discussing the most promising data collection strategies to meet the above criteria. Group discussions surfaced the following, noting that these strategies are not mutually exclusive:

• Adding questions or modules to existing federal survey efforts. This requires an assessment of what data is currently collected, what gaps exist, and the feasibility of adding questions to existing surveys (and which survey). For example: the ACS already asks whether someone has moved recently. Could a question be added to ask, “what was the reason for your move?” Questions can also be added relatively quickly to the Household Pulse Survey, which already asks about housing insecurity and reasons for moves. HUD is currently testing a housing insecurity module in AHS, which may prove instructive.
• Combining existing sources of survey data and administrative data to form a baseline of what we know at the national, state, and local levels. Administrative data could come from courts (where available), public entities with relevant data on moves (e.g., public electric and gas utility companies, FEMA for post-disaster moves), private companies like Attom Data and Corelogic that collect and sell eviction and foreclosure data, and existing public databases like HMDA. This approach would allow us to triangulate data sources to validate and fill gaps on transient populations who might not be reached with a survey. This approach could also be helpful in developing estimates for smaller areas or with greater frequency, while keeping data collection costs feasible. It may also provide a starting point, and surface what we already know and what we don’t know.
• Mandate that housing providers report housing loss data to a federal agency (similar to the way banks are mandated to provide mortgage data to CFPB). Under this model, data would be collected from large housing providers, as opposed to the households experiencing loss.
• Query public entities that might have relevant data on moves (e.g., public electric and gas utility companies) and ask what data is collected and whether it can be shared.
• Pilot an effort to establish a housing loss in one state or metro area, and use the lessons learned as a case study for scaling.

Attendees discussed three options, each with its own advantages and disadvantages.

Publicly led and funded: The federal government stewards a robust survey infrastructure and has significant survey expertise, including but not exclusively within the Census Bureau. A federal survey can be administered at scale and may be the most trusted and sustainable. Such an effort may require a Congressional mandate, though some executive action is possible within existing authority and appropriations. A disadvantage to this approach is that it would be harder for non-federal entities to shape data collection strategy or have oversight.

Privately led and funded: A private effort can be stood up more quickly, and will have fewer restrictions and more flexibility in reaching hard-to-reach and transient populations. A private effort is dependent on private funding, and would likely be smaller in scale. However, it could build the case for public investment. A promising precedent is the FDIC unbanked and underbanked survey, which grew out of a smaller survey effort led by a non-profit organization.

Private-public partnership: This could present a path for combining federal administrative data with private survey data (e.g., NORC’s Medicare Current Beneficiary Survey).