2025 Gateway Cities Housing Monitor | Chapter 1: Housing Production and Supply

These data show the absolute net change in housing units in Gateway Cities and their suburbs from 2022 to 2024, as well as the change relative to the size of the housing stock in 2022. You can see which cities are adding the most new units and which cities are increasing their housing supply at the fastest pace.  

Tracking the net change in housing units shows whether Gateway Cities are producing homes at the scale needed to ease housing pressures. The inaugural Housing Monitor found that they must roughly double production compared to the prior decade to keep up with demand. Sustained production above 2021 and 2022 levels signals that recent policy and funding interventions are working, while slower gains point to the need for stronger action to prevent worsening affordability challenges. 

To estimate production, we track the net change in residential addresses using records from the US Postal Service (USPS) aggregated by the US Department of Housing and Urban Development (HUD). While not a perfect measure of production, these data provide a more timely indication of how the housing stock is changing than do sources such as the US Census Bureau. To calculate the number of residential addresses, we deduct “No-Stat” addresses from the total residential address count. “No-Stat” addresses include buildings that are abandoned, or under construction and not yet ready to be occupied, as well as PO boxes registered with the Move to Competitive Street Addressing program. 

These data show how Gateway Cities and their suburbs are measuring up against the housing growth targets introduced in the inaugural Gateway Cities Housing Monitor. In 2024, the MassINC Policy Center estimated that the 26 Gateway Cities together would need to add 83,000 units to bring the housing market into balance. This target assumes 5 percent growth in the number of households (referred to here as population growth for simplicity) across Gateway Cities. This population figure was grounded in the assumption that all Gateway Cities are working to grow their populations along with their economies. It was also based on historical growth trends; the number of households in Gateway Cities increased by 8 percent over the previous 10 years. 

Measure 2A shows the cumulative progress that Gateway Cities have made toward the full 83,000-unit target since 2022. To gauge progress if household growth stalls due to more restrictive immigration policies and/or a slowing economy, Measure 2B presents a no growth scenario, which shows the unmet housing needs if existing population levels (as of 2022) continue unchanged.

Measures 2B and 2C follow progress at the city level. The 100-percent threshold indicates being on pace to reach the 2032 target. The dots show where each city stands relative to its target as of 2024.  

Tracking housing production against clear targets makes it possible to hold ourselves accountable for keeping pace with demand. When supply falls short of demand, rents and home prices rise, families face greater housing instability, and Gateway Cities risk losing momentum in revitalizing their downtowns and neighborhoods. Regular measurement provides an early warning system, showing whether we are on pace or if greater effort is needed to keep housing affordable and communities thriving. 

Of the 83,000 housing units Gateway Cities need to add over ten years to catch up to housing demand, about 8,000 units are needed to replace housing that is inevitably lost over time, whether due to fire, demolition, or other causes. Therefore, on net, the overall Gateway City housing supply needs to increase by about 75,000 units. Without population growth, this drops to just 36,000 units. Averaged out over time, this equates to 7,500 new homes per year on the high end or about 3,600 units per year on the low end. More information about how targets were calculated can be found in the 2024 Gateway Cities Housing Monitor.

Drawing from assessors’ records, these data show the size of new residential buildings and whether they were built from the ground up or by converting former mills, schools, churches, office buildings and other older structures. Together, these metrics reveal not just the scale of housing growth, but the physical forms and development strategies communities are relying on to expand their housing stock. 

The composition of new housing shapes the long-term trajectory of neighborhoods. Large-scale multifamily developments play a vital role by adding significant numbers of units, especially in locations that can support higher density. At the same time, “missing middle” structures with four to eight units provide a human-scaled transition between high-density centers and single-family areas, support walkable infill, and have historically offered a pathway to wealth-building for middle-income households. The balance between adaptive reuse and new construction is important to keep track of. Reuse projects can address blight, preserve historic character, and sometimes deliver housing more cost-effectively than new construction, but they face unique design and financing hurdles. As the most feasible reuse opportunities are absorbed, reliance on ground-up construction is likely to grow, raising new financial and regulatory challenges. 

These estimates were generated from municipal assessors data compiled by MassGIS. Annual reporting is encouraged but not consistent for all cities from year to year. Categories for building types are determined by available land use codes. Communities vary in how they apply land use codes, and many do not report unit counts. When unit counts are missing, they are estimated by dividing the square footage of the new building by the average residential square footage per unit of the existing housing stock. For FY23 to FY25, data were available for all Gateway Cities except Everett, New Bedford, and Revere. These gaps and inconsistencies mean that results are best interpreted as broad patterns in building and construction type rather than precise counts. 

This measure uses assessors’ records to identify the share of new housing located within half a mile of Amtrak and Massachusetts Bay Transportation Authority (MBTA) Commuter Rail stations. It focuses specifically on regional rail, which—unlike subway service in Boston—represents the primary rail service available to most Gateway Cities.

This level of detail provides a smart-growth lens on Gateway City housing production, showing whether growth is occurring in places where residents can best take advantage of transit and existing public infrastructure. Transit and housing thrive when they grow together. Concentrating new homes near rail stations builds walkable neighborhoods, expands access to jobs, and strengthens ridership. In turn, higher ridership justifies greater service, which further increases the value of living near transit. Tracking this measure shows whether new housing production is leveraging these intercity transit investments and helping to set off the mutually reinforcing cycle of stronger service and stronger housing demand. 

These estimates draw from municipal assessors’ data compiled by MassGIS. Annual reporting is encouraged but not consistent for all cities from year to year. For FY23 to FY25, data were available for all Gateway Cities except Everett, New Bedford, and Revere. These gaps and inconsistencies mean that results are best interpreted as broad patterns in building and construction type rather than precise counts. 

Permitting data show the number of new residential buildings authorized each year, distinguishing between smaller one- and two-family homes and larger three-or-more-unit buildings. Unlike address data, permits do not track net changes in housing units, since permitted projects are not always constructed and permit figures do not account for demolitions or the total number of units inside each building. Instead, they highlight patterns in the type and scale of new construction moving through the pipeline. 

While address data show how the housing stock has already changed, permitting data provide a forward-looking view. Because homes must be permitted before they can be built, permits serve as a leading indicator of both the overall volume of construction on the horizon and of the types of housing proposed—from smaller one- and two-family homes to larger multifamily developments. These trends reflect not only broader market conditions but also municipal choices about whether to allow more multifamily growth or rely primarily on single-family homes. 

The permitting figures reported here come from the US Census Bureau’s Building Permits Survey, which relies on municipalities to voluntarily report the number of permits they issue. Participation is inconsistent, with some cities providing complete data and others not. To fill these gaps, the Census Bureau imputes missing values using statistical methods. In 2024, data were imputed for 8 of the 26 Gateway Cities. While this means city-level figures may not always be precise, the approach is applied consistently enough over time that the trends remain meaningful. 

Units were estimated based on average per-building unit counts for cities reporting both the number of permits and the total number of units by building type. 

Drawing on several data sources, this indicator provides vacancy rates in both the rental and for-sale market, as well as in the aggregate across both types of housing. The United States Postal Service (USPS) collects quarterly information on whether an address is vacant based on reports from postal carriers. These USPS data provide near real-time monitoring of supply constraints, but they treat units as simply occupied or vacant. This means the data cannot distinguish between homes that are available for purchase and those that are available to rent.  

To capture these submarkets, we use the American Community Survey (ACS). While ACS estimates are less timely, they separate for-rent and for-sale vacancies. They also enable researchers to focus only on on-market housing, excluding units that are not occupied for other reasons (e.g., due to blight or because they are held for seasonal use). This makes the ACS a more precise measure of the vacancies that directly affect housing supply and price pressures. 

The natural vacancy rate is the share of housing units that must remain empty at any given time for the market to function smoothly, without persistent upward or downward pressure on prices. Some level of vacancy is necessary—households need time to move, landlords need time to prepare units, and buyers often sell one home before purchasing another. When rates fall below the natural vacancy rate, the competition for available units intensifies, and prices tend to rise. 

Economists generally use a benchmark of 5 percent for blended vacancy rates, which combine both rental and ownership units as in the USPS data. More detailed research by Harvard’s Joint Center for Housing Studies finds that the equilibrium differs by tenure; it is about 7.4 percent for rental housing, where turnover is high, and just 1.5 percent for ownership housing, where households typically stay put for longer periods. Taken together, these thresholds provide the benchmarks we use to evaluate vacancy trends in Gateway Cities. 

HUD Aggregated USPS Administrative Data: The vacancy counts reflect postal operations, not housing market activity. As USPS describes it, an address is either occupied and requires mail service or vacant and does not. This makes the measure timely but imperfect for capturing whether homes are actually available for occupancy. 

No-Stat addresses: The USPS also tracks “No-Stat” addresses, which include units under construction, demolished or condemned properties, and certain rural delivery points. These are excluded from the vacancy counts used here. 

ACS estimates: The American Community Survey (ACS) produces both 1-year and 5-year estimates, but 1-year results are limited to geographies with populations above 65,000. Because most Gateway Cities and all sub-city geographies fall below this threshold, we rely on 5-year estimates. As a result, the data smooth over short-term fluctuations, and small changes between periods may not be statistically meaningful.