Skip to main content

Mapping Neighborhood Change Using Near Real Time Administrative Data


Keywords: Data, Neighborhood Change Mapping Tool, Neighborhood Development, Tenant-based Voucher, American Community Survey, HUD Tenant Data, Natural Disaster, USPS, Public Housing Agency

Spotlight on PD&R Data
HUD USER Home > PD&R Edge Home > Spotlight on PD&R Data

Mapping Neighborhood Change Using Near Real Time Administrative Data

Alexander Din, Social Science Analyst, Office of Policy Development and Research
Todd Richardson, General Deputy Assistant Secretary, Office of Policy Development and Research

Neighborhood Change

In late 2023, HUD’s Office of Policy Development and Research (PD&R) updated the functionality of the U.S. Postal Service (USPS) data that it had been collecting for more than 15 years. Using the new Neighborhood Change Mapping Tool, users can now select one or more census tracts to get USPS administrative address data, HUD tenant data, and select variables from the decennial census and American Community Survey (ACS). Our agreement with the USPS allows nonprofits and government agencies to use these data to support their research and planning. Qualifying organizations can apply for access on

This tool offers numerous benefits and improvements over PD&R's previous data-sharing efforts:

  • The data are much more accessible. Previous data extracts were snapshots compiled quarterly or annually and required users to process many tables and analyze the data to generate insights. This point-and-click web map automates this process, allowing users to perform their analyses more quickly. Users no longer need to compile their own data, although other data options are still available.

  • The map tool is updated quarterly with data from the most recent quarter. We already have uploaded data from 2024-Q1, and we plan to update the map with 2024-Q2 data in July 2024.

  • The administrative data consist of 100 percent counts rather than survey data.

  • Because the map tool relies on point-level quarterly data, we have harmonized the data to 2020 census tracts, providing users a quarterly longitudinal dataset dating back to 2008.

Uses for the Data

The map tool is likely to be most useful for those with local knowledge of their study area. This tool can show neighborhood change with relatively current and specific data — for example, how many more units are under construction in the neighborhood? How many more households have moved into the neighborhood? How many units have become vacant?

Here are some ways in which local governments and nonprofits can use the map tool:

  • Have you made a land use policy change? These data might help you measure its impact.

  • Do you have a policy that requires developers to set aside a percentage of their projects' units as affordable housing? These data can show whether an increase in the number of housing units in a neighborhood also results in an increase in tenant-based voucher leasing.

  • Is a neighborhood gentrifying? A stable or increasing count of active addresses and a decline in voucher tenants may indicate gentrification.

  • How much housing in a neighborhood has been lost over time to decline, and how many vacant houses are there? Examine the change in active addresses for a decline in households, vacant housing counts for likely standing homes that are vacant, and the No-Stats data for lots that likely were left behind by demolition. An increase in tenant-based voucher leasing could mean that falling rents are making units more accessible to low-income tenants; a decrease in tenant-based voucher leasing could indicate a decline in housing quality.

  • Has your area experienced a disaster that displaced residents? This map tool can provide a count of destroyed homes and measure housing occupancy rates as the area recovers.

Measuring Post-Disaster Change

One of the powerful uses of the map tool is to monitor neighborhood change following a natural disaster. Map users can now upload a GeoJSON file, CSV file, or a compressed shapefile (and associated files) to create an overlay of a disaster area, which typically does not follow census tract boundaries. The exhibit below shows the August 2023 wildfire in Lahaina, Hawaii. The first image shows the overlay of the Lahaina Wildfire and three selected census tracts. The second image shows a stacked bar chart of administrative address data from USPS for the three selected tracts (highlighted in teal) from 2008-Q1 through 2024-Q1 — 65 quarters worth of data!

Map depicting an overlay of the Lahaina Wildfire and three selected census tracts.

A stacked bar chart of administrative address data from USPS for the three selected tracts from 2008-Q1 through 2024-Q1.

The number of active residential addresses grew slightly each quarter, from 2,759 in 2008-Q1 to 3,327 in 2023-Q3. After the wildfire, the number of active residential addresses fell to 1,244 in 2023-Q4. The wildfire has had a clear effect in the number of residential addresses actively receiving mail, our proxy for occupied housing units. There is also a large jump in the number of not-a-statistic, or No-Stat, residential addresses. This trend is similar to the one observed following the Camp Fire in Paradise, California. You will also find a similar pattern when looking at business addresses in Lahaina.

If you were to look only at census tract 15009031404, you would see the number of active residential addresses drop from 1,415 in 2023-Q3 to 17 in 2023-Q4, a 98.8 percent decrease. This number rises to 174 in 2024-Q1, which still represents a reduction of 87.7 percent for our proxy of occupied housing units.

You may be looking at the graph and wondering, “What is a No-Stat?” or “Why did the number of No-Stat addresses increase dramatically in 2011?” USPS has two classifications for unoccupied addresses:

  • Vacant addresses are addresses where no one has collected mail for at least 90 days. For this map tool, we have separated vacant addresses into two subcategories: a short-term vacant category for addresses that have been vacant for less than six months and may represent a more natural turnover process, and a long-term vacant category for addresses that have been vacant for six months or longer and may reflect an issue in the local housing market.

  • Not-a-statistic, or No-Stat addresses, are addresses that may be vacant (“No-Stat” indicating vacancy may be a more likely possibility in USPS-defined rural areas), demolished, merged with another address, under construction, or otherwise undeliverable. USPS does not detail why a particular address might be listed as No-Stat, although some published guidance instructs letter carriers to delete an address if it is undeliverable following a natural disaster.

In addition, in the early 2010s, USPS began its Move to Competitive Street Addressing program, which allowed USPS customers to register their PO Box as a street address to receive packages and deliveries from private carriers such as UPS and FedEx that require a street address for delivery. PO Boxes in this program are recorded as No-Stats, hence the sharp increase in No-Stats in 2011 for census tracts with a post office.

Growth in an Urban Neighborhood

The next example shows the Union Market neighborhood, just a short walk from the U.S. Capitol building in Washington, DC. For background, Washington, DC, has been growing the city by constructing new neighborhoods out of former industrial or warehouse areas. Most new housing construction in the District has been concentrated in a few neighborhoods in the central city. In 2008-Q1, the two census tracts (highlighted in the below image) that compose the neighborhood had 1,614 residential addresses actively receiving mail (our proxy for occupancy). In the early 2010s, developers began converting parking lots and warehouses in the neighborhood into housing. By 2024-Q4, Union Market had 8,583 occupied households, with more housing planned.

The stacked bar chart below shows the growth of active addresses over time and the growth in No-Stat addresses starting in 2016. Unlike the No-Stat addresses in the previous example, which occurred after a natural disaster, these No-Stat addresses likely were addresses under construction that eventually became active addresses.

Map depicting the Union Market neighborhood in Washington, D.C.

A stacked bar chart depicting the growth of active addresses over time and the growth in No-Stat addresses starting in 2016.

How accurate are these data?

A common question we receive is, "How accurate are the USPS data?" To answer this question, we provide 2010 and 2020 decennial census data on housing units to serve as a benchmark. Census data are useful because they are both sufficiently granular yet available nationwide, and they exist in multiple periods. Because discrepancies exist in the ways the U.S. Census Bureau and USPS define vacant housing, which result in different total housing unit counts, we recommend comparing occupied housing units from the decennial census with active residential addresses from USPS as the best two data points to compare.

Because the USPS data are complete as of the close of the quarter (March 31) and the decennial census data are complete as of April 1, we use 2010-Q1 and 2020-Q1 to compare the two data points. Examining the two census tracts in the Union Market neighborhood, we find that the USPS and decennial census data in both periods are close to each other. In 2010-Q1, USPS and the U.S. Census Bureau determined that 1,804 and 2,095 occupied housing units, respectively, were in the Union Market neighborhood. By 2020-Q1, those counts rose to 4,558 and 4,611 from USPS and the U.S. Census Bureau, respectively. The neighborhood has nearly doubled the number of occupied housing units in the past four years, with the USPS counting 8,583 active residential units as of 2024-Q1. If the U.S. Census Bureau conducted its decennial census again using a 2024-Q1 count of USPS active residential addresses, we’d expect the Census Bureau to find approximately 8,484 occupied housing units. We’ll revisit these estimates prior to the 2030 decennial census to revise these estimates.

Line graph depicting the growth of active residential addresses from USPS and Occupied Housing Units (count).

Occupied Housing Unit Counts

























What other features does the tool offer?

One of the most significant additions to the tool has been longitudinal, geographically harmonized HUD tenant data. Map users can click on any census tract and receive longitudinal information about various HUD rental assistance programs. The map tool draws from PD&R's quarterly extracts from HUD's Public and Indian Housing Information Center and Tenant Rental Assistance Certification System, which are 18-month snapshots of active recertifications, and new certifications submitted to HUD.

The image below compares the number of tenant-based voucher (TBV) households with the number of active residential units. At the beginning of the period, TBV households made up approximately 7 percent of the neighborhood’s occupied households. By the end of the period, TBV households dropped to slightly more than 2 percent of total households, approximately the national average for neighborhoods, but the count increased by two-thirds, from 111 to 185. The data indicate that the TBV households were able to grow with the neighborhood; a particularly important finding because most of the growth in residential units has consisted of newly constructed luxury apartments.

A line graph comparing the number of tenant-based voucher (TBV) households with the number of active residential units.

Map tool users can investigate HUD tenant data on the following rental subsidy programs:

  • Pubic and Indian Housing

    • Public Housing

    • Tenant-Based Vouchers

    • Project-Based Vouchers

  • Multifamily Housing

    • Project-Based Section 8 (PBS8)

    • Section 202 (Supportive Housing for the Elderly)

    • Section 811 (Supportive Housing for Persons With Disabilities)

    • Other Multifamily programs

These administrative data are based on data submissions from public housing agencies and owners of multifamily assisted housing on two HUD forms: HUD-50058 and HUD-50059. The data represent the count of active recertifications in an 18-month period before the end of the quarter. Some locations show surprising reductions and then increases from quarter to quarter that likely are due to data submission issues rather than an actual change in program participation; in these situations, the quarterly trend data can help users determine the difference. In addition, to protect privacy, if between one and 11 households participate in a program, we code the count as 11 households in that program.

What other information can the web tool provide?

Although PD&R doesn't plan to compile every possible variable, we have added additional data to help users better understand neighborhood change. In addition to housing data from the two recent decennial censuses, PD&R also has included population data from the decennial censuses, including both the count of a racial/ethnic group and its share of the population. We have also employed Simpson's Diversity Index, a common measure used in neighborhood change research.

The map tool includes data on race and ethnicity from the 2010 and 2020 decennial censuses. Examining the share of population that is Black, we find that the Union Market neighborhood dropped from 51.7 percent Black in 2010 to 32.5 percent Black by 2020. This change is unsurprising given the history of gentrification in Washington, DC. However, when we look at the count of Black population in the neighborhood, we see the number of Black residents rose from 3,051 in 2010 to 3,454 in 2020.

 A line graph depicting the growth of active residential addresses with the population of African American or Black Non-Hispanic (rate).

A line graph depicting the growth of active residential addresses with the population of African American or Black Non-Hispanic (count).

The map also includes select ACS variables as well as data from three surveys:

  • The 2008 – 2012 ACS, which estimates neighborhood conditions in 2010.

  • The 2013 – 2017 ACS, which estimates neighborhood conditions in 2015.

  • The 2018 – 2022 ACS, which estimates neighborhood conditions in 2020.

We use data from three separate ACS surveys to ensure that map users who are comparing changes in neighborhood characteristics know that these data are drawn from different rolling samples. We also provide the margins of error for the ACS data.

Our data draw from literature on neighborhood change and include data on the share of the population over age 25 that is college educated, the share of population that is foreign born, the share of households that are at or below the federal poverty level, and the share of population aged five or older that speaks English at home.


2008–2012 ACS

2013–2017 ACS

2018–2022 ACS




College Educated




Foreign Born




Poverty Rate




Speaks English at Home




A line graph depicting the growth of active residential addresses with the percentage of college educated population.

What plans do you have for the future?

First, we want to hear from you. Although these data are insightful, your knowledge of local geography is what makes this tool powerful.

Second, processing and preparing these data for use has involved considerable effort. We are still reviewing and revising the Neighborhood Change Mapping Tool. To continue improving the map tool, we want to know how you are using the tool and how it could better help you get the information you need.

We welcome your suggestions for data to include in future iterations of the map tool. We are especially interested in additional administrative data that are granular both in space (e.g., point level) and time (e.g., quarterly).

We plan to release another PD&R Edge article comparing active residential addresses and occupied housing units across the 50 states and the District of Columbia. Want another analysis? Please reach out to Alex Din at, tell him how you’re using the data, and ask for another analysis.


Published Date: 11 June 2024

The contents of this article are the views of the author(s) and do not necessarily reflect the views or policies of the U.S. Department of Housing and Urban Development or the U.S. Government.