An ambitious new state law championed by Maryland Governor Wes Moore will allow denser multifamily housing within three quarters of a mile of the state’s many rail transit stations. The law also includes an inclusionary zoning (IZ) provision, requiring 15% of new units be set aside as affordable.

However, like many local IZ ordinances, it defines “affordable” as available to people making up to 60% of the area median income (AMI), adjusted for family size. To occupy such a unit, applicants must document their eligibility, which takes time, and then pay either 30% of their income or the listed rent, whichever is lower. 

But does Maryland need mandates to deliver new housing at this price point? In the Baltimore and DC metro areas, where most Marylanders live, over a third of recently built rental housing units are rented at rates that qualify as affordable. To be sure, that third includes some IZ and directly subsidized units. But most new rentals are market rate, and a solid proportion of those turn out to be affordable to low-income households.

Incomes and outcomes

But this isn’t really a story about the power of the market. It’s a story about area median income. In metropolitan Baltimore, a family of four making $73,000 in 2024 qualifies for 60% AMI affordable housing, where it would pay $1,825 per month for rent, utilities included. A third of new market-rate three-bedroom units in Baltimore are rented at around that level.

Baltimore is typical, as it turns out. In most U.S. metro areas, a substantial share of rentals constructed since 2010 were, in 2021 and 2022, affordable at 60% of AMI. The below map shows this share across regions. You can also check out maps showing rentals affordable at 80% and 120% of AMI.

It might be surprising that roughly a third of new rentals are within reach of low income renters in expensive metropolitan areas like Greater Seattle and Austin. The reason is that these metros, like Greater Baltimore, have high median family incomes. Great jobs mean great incomes for many families in those regions. In metro Seattle, median family income is $140,700. By standard administrative definitions, that means a unit renting at $2,100 per month is affordable to a family of four that earns 60% of AMI.

In some metro areas, such as those of Raleigh and St. Louis, the majority of recently-built rentals are affordable to families earning 60% of AMI. In the metro area of slow-growing St. Louis, extremely low rents on a fifth of recently-built rentals suggest that a substantial share of new units are deed-restricted affordable housing. 

At the opposite extreme are Florida metros, including Greater Orlando, where just 12% of recently-built rentals are affordable at 60% of AMI. Taking into account that this includes subsidized housing and some statistical noise, it seems that very little Florida development is affordable at market rates. One reason is that rent is somewhat higher in Orlando than Raleigh. A less obvious reason is that retirees are included in AMI: Florida has many homeowning seniors with low current incomes. As a result, all AMI measures are low relative to the distribution of renter incomes.

But these comparisons don’t really tell us how well Orlando and Raleigh’s markets are serving renters. Rather, they tell us how those two markets look when filtered through the administrative lens of AMI.

Who does IZ serve?

None of this means that market-rate new construction is affordable to poor people – it just means that 60% of AMI is not necessarily poor. And in most of the US, households earning 60% of AMI are well served by the market.

Of course, the filtering of older housing units is a big part of that. But, as this analysis shows, a substantial share of new housing in most metropolitan areas will be within range for households with decent, below-average incomes.

That’s great. But it also means that IZ programs with high cutoffs are offering Potemkin affordability. Landlords like it because 60% AMI tenants are within their normal market range and have fewer credit and income challenges than poor tenants. Politicians like it because they can claim credit for “affordable” units without making hard budgetary choices. But effectively providing affordable housing for the poorest families requires much larger tax abatements or subsidies. 

Potemkin affordability is popular: a 2016 report found that just 4% of IZ units created in California were affordable at 30% AMI or less. Another quarter went to people in the 30 to 50% AMI band; the rest were for people with jobs such as “school teacher” and “civil engineer.”

But Potemkin affordability is not without costs. Yuri Geylik, co-founder of the affordable housing consultancy MGNY, estimates that rent in deed-restricted units needs to be discounted 10-15% below market rate to attract tenants. In other words, even in the tight New York City rental market, a $200 per month rent discount is not worth the hassle and delay of qualifying for an affordable unit to middle-class tenants. And that doesn’t even count the administrative costs to the property manager.

At the opposite end of Interstate 80, property managers have even greater difficulty filling units reserved for middle-income tenants. J.K. Dineen recently reported that 85% of new San Francisco apartments reserved for 100-150% AMI familiesin that category are sitting vacant because proving eligibility is a “tortured and time-consuming process with as much paperwork as it would take to buy a home.”

Putting assistance into focus

Taken together with previous research on inclusionary zoning, we can draw a few clear takeaways from this analysis for affordability policy:

First, targeting affordable units to families at or above 60% of AMI may indirectly hurt those in much greater need. Not only does a high-threshold IZ policy direct private subsidies toward those with above-poverty incomes, but it may also discourage construction, putting moderate-income and poor people in competition for the same limited pool of existing rentals.

When inclusionary zoning policies discourage market rate production, as Shane Phillips estimates, cities lose a large number of new units affordable to households earning 50 to 100% of AMI. And poorly-targeted IZ can use up political capital and attention that would be better focused on those who cannot afford market-rate housing.

Second, assistance programs should be fully funded via tax abatements or vouchers. Although outside the scope of this analysis, a strong consensus is emerging that unfunded IZ does more harm than good. Funded IZ can still pose administrative difficulties, but it is much more likely to result in more, rather than less, housing for people with below-median incomes. Funded IZ policies in Baltimore and Portland provide possible models.

Third, policies that unlock lower-cost housing types will reach deep into the income distribution. In every metro area, at least some new housing is affordable to low-income families. No-frills apartments and build-to-rent subdivisions are good candidates for housing that is attainable to low-income families.

Conclusion

Usually, the IZ debate is over whether IZ discourages housing production so much that the loss of market-rate units is not worth the gain in affordable units. But that question presupposes that market-rate units are out of reach for the tenants to whom the affordable units are targeted.

Often, that’s not the case: many new, market-rate apartments are affordable to the very families that IZ programs are intended to help. Policymakers should respond to this finding by narrowing the focus of subsidies and accelerating the production of market-rate housing.

Salim Furth is a Senior Research Fellow at the Mercatus Center at George Mason University. He holds a Ph.D. in Economics from the University of Rochester.

Appendix: Methods

I used 2021 and 2022 American Community Survey (ACS) data (via IPUMS) on incomes and rents, and the family median income metric that grounds HUD’s AMI calculations. None of the data are perfect; actual incomes are higher than those reported to ACS survey-takers, for example. My AMI calculations do not perfectly match HUDs, primarily because HUD divides some metro areas, especially in the Northeast, into bespoke sub-regions.

The key choice in my analysis is how to assign housing units to family size. The larger I set the “feasible family size” for each unit, the more likely it will be marked “affordable”, because HUD’s affordability formula sets a higher income, and thus higher potential rent, for larger families.

The median American home has an equal number of bedrooms and occupants. But there’s plenty of variation. For each rented home in the data, I estimated “feasible family size” based on the number of bedrooms, rooms, and occupants reported to the ACS:

FEASIBLE FAMILY SIZE = min (rooms, 2*bedrooms, max(bedrooms+1, observed occupants))

To this formula, I made one exception: 1 bedroom, 2 room apartments with 1 observed occupant are assigned a feasible family size of 1, not 2. This exception reduces the number of rental units that I mark as affordable.

The map displays only metropolitan areas in which there are at least 150 observations of renters in units built since 2010.

This analysis ignores most distinctions among new rental units, especially location and amenities. For example, 1-bedroom apartments in Boston’s core neighborhoods are 30 percent more expensive than in the affluent suburbs along Route 128.