This dashboard projects the impact of legalizing starter homes nationally and in the 50 states by examining new single-family homes built in new subdivisions from 2015–2024.
The underlying data cover nearly all assessed residential properties nationwide. This analysis focuses on about 4.6 million single-family homes built from 2015–2024 in new subdivisions.
Two findings are especially important.
First, the periodic table shows how land value and lot size interact. Where smaller lots are legal, builders can use land more efficiently. The most attainable new homes tend to be built where smaller lots are combined with less expensive land. In many states, these homes qualify as starter homes under our definition, which is based on the home-value-to-county-income ratio. The pattern is striking: lot size, together with land price, is among the most important factors shaping the affordability of newly built homes.
Also of note, living area is highly correlated with lot size, which further helps explain how smaller lots can support the production of more starter homes. Finally, even starter homes on the smallest lots average roughly 1,400 to 1,500 square feet of living area, generally sufficient for three bedrooms. These are not tiny homes.
Second, starter homes are already being built in many areas—but not in the quantities needed. The data show that it is possible to build starter homes today; the question is whether land-use rules and other regulations, including aesthetic requirements, allow enough of them to be built. Based on case studies and conversations with home builders, we expect builders would produce more starter homes if the rules allowed them to do so.
The dashboard estimates how many more homes could be produced if states or local jurisdictions allowed new subdivisions to use smaller lots by right and without provisions that materially limit feasibility. Results can be viewed nationally, by state, and, for many measures, by county. The dashboard also makes the policy tradeoffs transparent. Users can test how the results change when reform applies only to larger subdivisions, only to counties above a selected population threshold, or subject to different maximum allowed minimum lot sizes and market-response assumptions.
The results point to a clear policy opportunity for state legislatures: removing barriers to smaller-lot development in new subdivision could substantially increase starter-home production. To put the potential scale in perspective, under the best-case scenario, we estimate that roughly 3.4 million starter homes could have been built nationally from 2015–2024, compared with about 840,000 actually built. The implication is straightforward: changing these rules now can help ensure that the next decade does not repeat the missed opportunities of the last one.
The potential for policy missteps is also substantial. If the maximum allowed minimum lot size is set at 3,000 square feet rather than 1,000 square feet, estimated U.S. starter-home potential falls from 2.5 million to 1.3 million, a 48% reduction.
Here is a short overview of each tab in the tool.
This tab shows how new homes are distributed across the home-value-to-county-income spectrum and how many qualify as starter homes under different affordability thresholds, with 4x being the default. It answers a central question: How much of today’s new-home production reaches starter-home price points, and how much more could be built under smaller-lot scenarios?
This tab shows the distribution of new homes by home value relative to county median income and by lot size. It also models how policy changes would shift these distributions.
This tab summarizes the main modeled effects of lot-size reform for the selected geography and scenario. It compares actual new-home production with modeled production under different policy settings and filters. It is the quickest way to see the overall scale of the modeled impact on total homes and starter homes, and how those results change with the assumed lot-size shift and maximum allowed minimum lot size.
This tab shows how land value and lot size interact to shape home values and affordability. Homes are grouped into land-value deciles and lot-size bins and color-coded by home-value-to-county-income ratio. Home value-to-income ratio improves dramatically when smaller-sized homes are built on smaller-sized lots (columns to the left) in subdivisions on less expensive land (lower rows). Nationally, starter homes are generally built at the lower left of the Periodic Table.
Conversely, homes built on larger lots and more expensive land tend to produce larger and more expensive homes. Users can also view other key metrics, including home values, living area, bedrooms, and the number of homes in each cell.
This tab examines how lot size and land value relate to the size of the home itself. The scatterplot shows median gross living area across land-value deciles and lot-size bins. Importantly, smaller lots do not mean tiny homes: even starter homes on the smallest lots average roughly 1,400 to 1,500 square feet of living area, generally sufficient for three bedrooms.
This tab compares the estimated potential for additional new-home production across states under the selected reform scenario. It shows where lot-size reform could have the largest impact, both in absolute terms and relative to the existing housing stock, helping identify the states with the greatest opportunity to expand starter-home supply through land-use reform.
This tab aggregates the modeled results nationally and ranks states by their potential for additional starter-home production. Nine states account for 79% of the estimated starter-home potential, and these same states are also projected to experience some of the largest population gains through 2040. This helps show where state starter-home reform could have the greatest national impact.
This tab provides a more detailed look at how modeled production changes under different policy assumptions, including subdivision size, county-population thresholds, lot-size shifts, and maximum allowed minimum lot sizes. It allows readers to test alternative policy designs, assess how sensitive the results are to those choices, and see directly how each assumption changes projected home and starter-home counts.
This tab estimates how modeled increases in new-home production affect annual property-tax revenue at the state and county levels. By allowing more homes to be built on the same amount of land, smaller-lot reform can expand the local tax base even when the individual homes are less expensive.
This tab translates the analysis into a concise, plain-English summary and highlights the urgency of reform. It is designed to make the findings easy to communicate to policymakers, stakeholders, and the public without requiring them to work through the full dashboard. The order of the pitch is designed to elicit broad agreement presented to a legislator or other interested party, and at around the mid-point lead the listener to volunteer his or her starter home experience. The last portion attempts to “close” on the sale: we must end the ban on starter homes and it needs to be done at the state level.
What does this tab show?This tab shows how new homes are distributed across the home-value-to-county-income spectrum and how many qualify as starter homes under different affordability thresholds, with 4x being the default. It answers a central question: How much of today’s new-home production reaches starter-home price points, and how much more could be built under smaller-lot scenarios?
What does this tab show?This tab shows the distribution of new homes by home value relative to county median income and by lot size. It also models how policy changes would shift these distributions.
What does this tab show?This tab summarizes the main modeled effects of lot-size reform for the selected geography and scenario. It compares actual new-home production with modeled production under different policy settings and filters. It is the quickest way to see the overall scale of the modeled impact on total homes and starter homes, and how those results change with the assumed lot-size shift and maximum allowed minimum lot size.
| Maximum allowed minimum lot size ?The largest minimum lot-size requirement a jurisdiction may impose—in effect, the lower bound on permitted lot sizes. |
Total homesNumber of homes | Starter homesNumber of homes | Total homesHome-value-to-county-income ratio | Starter homesHome-value-to-county-income ratio | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Actual | Below potential | Somewhat below potential | Full potential | Actual | Below potential | Somewhat below potential | Full potential | Actual | Below potential | Somewhat below potential | Full potential | Actual | Below potential | Somewhat below potential | Full potential | |
| Maximum allowed minimum lot size ?The largest minimum lot-size requirement a jurisdiction may impose—in effect, the lower bound on permitted lot sizes. |
Total homesNumber of homes | Starter homesNumber of homes | Total homesHome-value-to-county-income ratio | Starter homesHome-value-to-county-income ratio | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Below potential | Somewhat below potential | Full potential | Below potential | Somewhat below potential | Full potential | Below potential | Somewhat below potential | Full potential | Below potential | Somewhat below potential | Full potential | |
What does this tab show?This tab estimates how modeled increases in new-home production affect annual property-tax revenue at the state and county levels. By allowing more homes to be built on the same amount of land, smaller-lot reform can expand the local tax base even when the individual homes are less expensive.
| Maximum allowed minimum lot size ?The largest minimum lot-size requirement a jurisdiction may impose—in effect, the lower bound on permitted lot sizes. |
Annual Property Tax | Annual Property Tax Increase | % Increase Annual Property Tax | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Actual | Below potential |
Somewhat below potential |
Full potential |
Below potential |
Somewhat below potential |
Full potential |
Below potential |
Somewhat below potential |
Full potential |
|
What does this tab show?This tab compares the estimated potential for additional new-home production across states under the selected reform scenario. It shows where lot-size reform could have the largest impact, both in absolute terms and relative to the existing housing stock, helping identify the states with the greatest opportunity to expand starter-home supply through land-use reform.
What does this tab show?This tab aggregates the modeled results nationally and ranks states by their potential for additional starter-home production. Nine states account for 79% of the estimated starter-home potential, and these same states are also projected to experience some of the largest population gains through 2040. This helps show where state starter-home reform could have the greatest national impact.
What does this tab show?This tab shows how land value and lot size interact to shape home values and affordability. Homes are grouped into land-value deciles and lot-size bins and color-coded by home-value-to-county-income ratio. Home value-to-income ratio improves dramatically when smaller-sized homes are built on smaller-sized lots (columns to the left) in subdivisions on less expensive land (lower rows). Nationally, starter homes are generally built at the lower left of the Periodic Table. Conversely, homes built on larger lots and more expensive land tend to produce larger and more expensive homes. Users can also view other key metrics, including home values, living area, bedrooms, and the number of homes in each cell.
What does this tab show?This tab examines how lot size and land value relate to the size of the home itself. The scatterplot shows median gross living area across land-value deciles and lot-size bins. Importantly, smaller lots do not mean tiny homes: even starter homes on the smallest lots average roughly 1,400 to 1,500 square feet of living area, generally sufficient for three bedrooms.
What does this tab show?This tab translates the analysis into a concise, plain-English summary and highlights the urgency of reform. It is designed to make the findings easy to communicate to policymakers, stakeholders, and the public without requiring them to work through the full dashboard. The order of the pitch is designed to elicit broad agreement presented to a legislator or other interested party, and at around the mid-point lead the listener to volunteer his or her starter home experience. The last portion attempts to “close” on the sale: we must end the ban on starter homes and it needs to be done at the state level.
What does this tab show?This tab provides a more detailed look at how modeled production changes under different policy assumptions, including subdivision size, county-population thresholds, lot-size shifts, and maximum allowed minimum lot sizes. It allows readers to test alternative policy designs, assess how sensitive the results are to those choices, and see directly how each assumption changes projected home and starter-home counts.
Methodology
New-home production data come from First American and reflect nationally aggregated public-record tax assessor data available as of July 2026 for virtually every U.S. county. The data cover nearly all assessed residential properties and include key property characteristics such as land-use code, year built, number of bedrooms, address, and gross living area (GLA). The analysis includes single-family homes built from 2015–2024 in new subdivisions.1 To improve the accuracy and consistency of lot-size measures, we replace assessor-reported lot sizes with parcel areas calculated from First American’s parcel-level shapefiles; the resulting lot-size measures are rounded to the nearest 10 square feet. Where bedroom counts are missing, we supplement them with First American Listings data when available. Current home values are measured using December 2025 automated valuation model (AVM) estimates from First American. Using a common valuation date allows homes built in different years to be compared on a consistent current-value basis.
Median household income, population totals, commute times, and housing-stock estimates—including total housing units and 1–4-unit housing stock plus mobile and manufactured homes—come from the 2020–2024 five-year American Community Survey (ACS) at the county level. Per ACS methodology, county median household income estimates from prior years are adjusted forward to 2024 levels. Estimated land value is derived from the AEI Housing Center’s land-value methodology described below.
For each newly built home, we calculate the ratio of its December 2025 AVM to the 2020–2024 ACS five-year median household income of the county in which the home is located. We refer to this as the home-value-to-county-income ratio, or P/I ratio.
Starter homes are defined using a selected threshold for home value relative to county median household income. This measure provides a simple, locally adjusted affordability benchmark by relating home values to the incomes of households in the same county. We use county median household income rather than metropolitan area median income because the goal is to assess whether homes are attainable for households already living in that county; metro-wide income measures can differ significantly as they can be affected by higher- or lower-income counties elsewhere in the metropolitan area. The dashboard allows thresholds of 4.0×, 4.5×, 5.0×, and 6.0× county median household income, with 4.0× used as the primary specification.
Unless otherwise noted, statewide medians are calculated directly from the underlying homes in the state. Values marked with an asterisk (*) are estimated as weighted medians of county-level medians, as described below.
For statewide summary measures that require a single income benchmark, the dashboard uses the median of county median household incomes weighted by the number of new homes built in each county. This is distinct from the ACS statewide median household income. A home's P/I ratio itself always uses the median household income of the county in which that home is located.
New single-family homes built from 2015–2024 are grouped into the following lot-size bins: less than 1,000; 1,000–1,199; 1,200–1,399; 1,400–1,999; 2,000–2,999; 3,000–3,999; 4,000–4,999; 5,000–5,999; 6,000–6,999; 7,000–7,999; 8,000–8,999; 9,000–9,999; 10,000–11,999; 12,000–14,999; 15,000–17,999; and 18,000–21,780 square feet.
The dashboard models three possible market responses: below potential, somewhat below potential, and full potential, with the last serving as the default scenario. The below-potential scenario assumes a uniform one-bin reduction in lot size, the somewhat-below-potential scenario assumes a two-bin reduction, and the full potential scenario assumes a three-bin reduction. This full potential scenario assumes smaller lots are allowed by right and does not include “poison pill” provisions that materially limit feasibility.
Under each market-response scenario, homes are reassigned to a smaller lot-size bin according to the selected shift. Homes already in the smallest lot-size category remain unchanged.
When a home is shifted to a smaller lot-size bin, its hypothetical lot size is set at the same relative position within the destination bin as its observed lot size occupied within the original bin.2 For example, a home on a 17,000-square-foot lot falls in the 15,000–17,999-square-foot bin, about two-thirds of the way through that range. If shifted down three bins, it moves to the 9,000–9,999-square-foot category and is assigned a hypothetical lot size of about 9,667 square feet. This percentile-preserving approach maintains differences in lot sizes within each bin rather than assigning every home in a destination category the same midpoint value.
Hypothetical home supply capacity is then calculated as the home’s observed lot size divided by its hypothetical lot size. In the example above, 17,000 square feet divided by 9,667 square feet yields a capacity of 1.76 homes, or 0.76 additional home relative to the observed one home. Because capacity can be fractional, these values are intended to be aggregated across homes and should not be interpreted as a literal number of homes built on an individual parcel.
The market-response scenarios are counterfactual simulations rather than forecasts of the exact number, size, or type of homes that builders would produce following reform. They use observed patterns of new construction and within-tract relationships among lot size, home size, and home value to estimate how allowing smaller lots could affect production capacity, home values, and living area.
The model holds the general geographic pattern of observed development constant and does not explicitly model changes in land prices, the location of future development, infrastructure costs, builder profitability, housing demand, permitting timelines, or other behavioral responses that could follow reform.
Moving a home to a smaller lot is assumed to affect both its market value and its GLA. These effects are estimated separately for each state using log-log regressions relating home value to lot size and GLA to lot size, with census-tract fixed effects.
The tract fixed effects mean that the estimated relationship is identified from differences among homes within the same census tract, rather than from differences across census tracts with different land values, amenities, or housing-market conditions. This approach is well suited to the hypothetical exercise because the market-response scenario changes the size of the lot while holding the property's general location constant.
Because both lot size and the outcome variable (AVM or GLA) are expressed in logarithms, the resulting coefficients are elasticities. For example, a lot-size elasticity of home value of 0.10 implies that a 10% smaller lot is associated with approximately a 1% lower home value, holding census tract constant.
The state-specific elasticities are then applied to each home's change from its observed lot size to its hypothetical lot size to estimate the corresponding hypothetical AVM and GLA. The home-value-to-county-income ratio is then recalculated using the hypothetical AVM and the same county median household income.
The dashboard also models policies that cap how large a jurisdiction may set its minimum lot-size requirement. Users may select no cap or a maximum allowable minimum lot size of 1,000; 1,200; 1,400; 2,000; 3,000; 4,000; and 5,000 square feet.
For example, under a 3,000-square-foot maximum allowed minimum lot size, the modeled shift cannot result in a hypothetical lot smaller than 3,000 square feet. A home may otherwise shift downward by the selected number of lot-size bins, but its hypothetical lot size cannot fall below the selected threshold. Homes already on lots at or below the selected threshold remain unchanged.
The dashboard also allows the market-response scenario to be limited to subdivisions meeting a selected minimum size. Available categories are No minimum (all subdivision sizes), 3+ acres, 5+ acres, and 10+ acres. The categories are cumulative: the 5+ acre category includes all subdivisions of at least 5 acres, while the 10+ acre category includes all subdivisions of at least 10 acres. When a subdivision-size category is selected, the market-response scenario is limited to homes in subdivisions meeting that threshold. Homes in smaller subdivisions remain at their actual observed values.
Subdivision size is estimated using two complementary methods: subdivision name within ZIP code and a geospatial nearest-neighbor match of homes built within ±1 year in the same census block group. In both cases, a subdivision must contain at least 10 homes to be included. See the AEI Housing Center New Residential Subdivision Classification Methodology for additional detail.
For each identified subdivision, the observed lot sizes of all included homes are summed to estimate total residential lot acreage. This total is then adjusted upward to approximate the gross land area of the subdivision, including land used for roads, drainage, retention ponds, and other common infrastructure. The adjustment factor varies by state to reflect differences in climate, topography, drainage requirements, and typical subdivision design.3
The dashboard also allows the market-response scenarios to be limited to counties above a selected population threshold. When a county-population cutoff is not engaged, statewide results use the exact statewide distributions and medians calculated from the underlying property-level observations.
If a county-population filter is applied, the tool automatically switches to a statewide analysis. The selected market shift is applied only to counties whose populations are at or above the cutoff. Counties below the cutoff remain at their actual observed values.
For these statewide market shift scenarios that depend on population cutoffs, home counts and AVM-to-income distributions are calculated by directly summing the applicable county-level results. Counties meeting the population cutoff contribute their shifted distributions, while counties below the cutoff contribute their observed actual distributions.
Because the population cutoff can create many combinations of shifted and unshifted counties, exact pooled statewide medians are not pre-calculated for every possible county-population cutoff scenario. Instead, counts and distributions are aggregated directly from county results, while statewide median lot size and GLA are estimated as weighted medians of county-level medians, using each county's observed number of homes built as the weight. These estimated medians are marked with an asterisk (*).
For consistency, the actual value shown alongside a county-population cutoff median is calculated using the same weighted-county-median method rather than the exact pooled statewide actual. These estimates are therefore intended primarily to measure the difference between the actual and scenario values rather than the absolute level of the median.
In county-population cutoff scenarios, starter-home median lot size and GLA are similarly estimated as weighted medians of county starter-home medians. The weights are based on each county's starter-home count under the applicable market-response scenario and selected starter-home threshold; the corresponding actual uses actual starter-home counts.
Estimated land value is derived from the AEI Housing Center’s land-value methodology, which separates a property’s AVM into estimated structure and land components. Structure value is estimated from construction costs and adjusted for depreciation and obsolescence; the remaining value is attributed to land. Property-level land-share estimates are partially smoothed using the census-tract median. See the AEI Housing Center land-value methodology for additional detail.
Each home is assigned the median estimated land value of pre-1990 homes within its 2020 census tract, using 2020 census tract boundaries. The measure reflects the estimated total land value of the typical pre-1990 single-family parcel in the tract, rather than land value per square foot. This is intentional: the analysis seeks to capture the cost of obtaining a buildable residential parcel under the prevailing parcel pattern and land-use regime, rather than normalizing away differences in parcel size.
This serves as a measure of underlying land-market conditions. Tracts without sufficient pre-1990 homes may not have an estimated land value; approximately 4% of tracts and 8% of homes lack a land-value estimate. For these tracts, land value is imputed using the five geographically nearest census tracts with observed land-value estimates, based on the distance between tract centroids. The neighboring tracts are weighted by the number of pre-1990 homes underlying their land-value estimates. We evaluate the imputation method using random 10% holdout samples in the states with the largest shares of census tracts missing a land-value estimate.
Estimated land value is divided into 10 equally sized state-specific deciles, from the lowest-value decile through the highest-value decile. Homes built from 2015–2024 are then grouped jointly by land-value decile and 14 lot-size bins to form the periodic table.4 The lot-size display can use either detailed bins below 3,000 square feet or a combined <3,000-square-foot bin. The combined view is useful for states with relatively few observations in the smaller lot-size categories, where pooling those bins reduces noise and makes the underlying pattern easier to interpret.
For states with at least 20,000 new homes in the periodic-table analysis, the state periodic table is shown first with the U.S. table below. For states below that threshold, the U.S. table is shown first, and the state table follows with a small-sample warning.
Within each land-value-decile × lot-size cell, the periodic table can display the median P/I ratio, median AVM, median county household income, median GLA, average number of bedrooms, or number of observed homes. Cells with fewer than 25 observations are suppressed.
For most metrics, the dashboard allows users to choose between a State-specific and a National color scale. Under the State-specific color scale, colors are based on the distribution of eligible displayed cells for the selected state and selected lot-size display. The state's 10th-percentile value serves as the lower endpoint of the scale and is shown using the darkest green, while the state's 90th-percentile value serves as the upper endpoint and is shown using the darkest red. Values below or above these endpoints receive the corresponding endpoint color. The midpoint of the color scale reflects the state's median value.
Under the National color scale, the same fixed national reference points are used for every state. This makes the colors directly comparable across states: a given value receives the same color regardless of which state is being viewed. The national reference distribution is calculated separately for each metric and does not change when the user switches between the detailed and combined lot-size displays.
For the P/I ratio, the National scale uses 3.0× county median household income as its lower reference point, the national median as its midpoint, and the national 90th percentile as its upper reference point. Values at or below 3.0× receive the lowest-cost color, while values at or above the national 90th percentile receive the highest-cost color. The United States periodic table uses this same fixed National scale.
Cell n-counts use geography-specific color scales rather than the common National scale. When n-count is selected, the National color-scale option is disabled; state tables are colored using the selected state’s n-count distribution, while the U.S. table is colored using the national n-count distribution.
We use taxing-jurisdiction-level property-tax rates from the AEI Property Tax Atlas, which calculates effective and new-construction property-tax rates for U.S. cities and unincorporated counties using public-record data covering approximately 84 million single-family homes. See the Property Tax Atlas methodology for more details.
For each observed home in the sample, estimated annual property-tax revenue is calculated by multiplying its December 2025 AVM by the applicable new-construction property-tax rate. Under each modeled market-response scenario, the hypothetical per-home AVM is multiplied by modeled housing capacity and the same local tax rate. These estimates are then aggregated to the county and state levels. The difference between estimated revenue under each market-response scenario and actual estimated revenue is reported as the modeled increase in annual property-tax revenue.
1 For Jefferson County, KY, the data are for 2013-2025. ↩
2 For the lowest bin, hypothetical lot sizes are bounded between 800 and <1,000 square feet. ↩
3 The factors generally fall into the following ranges: 1.23–1.28×: dry or relatively flat states where dedicated stormwater acreage tends to be limited; 1.30–1.33×: typical U.S. subdivision conditions; 1.35–1.39×: wetter states, areas with more substantial stormwater requirements, or less-efficient subdivision layouts; approximately 1.41×: Florida- or Louisiana-type environments where drainage and water management can consume a substantial share of subdivision acreage. ↩
4 These bins are: less than 1,000; 1,000–1,199; 1,200–1,399; 1,400–1,999; 2,000–2,999; 3,000–3,999; 4,000–4,999; 5,000–5,999; 6,000–6,999; 7,000–7,999; 8,000–8,999; 9,000–9,999; 10,000–14,999; and 15,000–21,780 square feet. ↩