Building the Skyline

The Birth & Growth of Manhattan's Skyscrapers

by Jason Barr 1 Comment

Why Has Gotham Been in a “Housing Emergency” for 80 Years? (Part I)

Jason M. Barr               January 22, 2025

If you google the definition of “emergency,” the search engine provides a response from the Oxford Languages Dictionary: “a serious, unexpected, and often dangerous situation requiring immediate action.” Merriam-Webster comes in with similar wording: “an unforeseen combination of circumstances or the resulting state that calls for immediate action.”

So clearly an emergency is sudden, unforeseen, and requires immediate corrective action to fix a dangerous situation. And yet paradoxically, for more than three-quarters of a century, New York’s housing market has been in a state of “emergency.”

In March of 2024, once again,

the New York City Council voted to declare that there is an ongoing housing emergency to extend the Rent Stabilization Law from expiring on April 1, 2024, to April 1, 2027. To maintain New York City’s rent stabilization laws pursuant to state law, the Council must determine whether there is an ongoing housing emergency every three years, defined as a vacancy rate of less than 5% of the City’s rental housing stock. Findings from the 2023 Housing and Vacancy Survey (HVS) show a citywide rental vacancy rate of 1.41%, the lowest this measurement has been since 1968, and a significant decline from 2021, when the net vacancy rate was 4.54%.

The Paradox

But if we take a step back, we can see that it’s impossible that there is an “emergency” both as New York defines it and as Webster and Oxford define it. Something cannot be an emergency if it’s predictable, ongoing, and with which people have lived for a very long time.

And yet since the early 1940s, when the language was introduced because of a true housing emergency caused by World War II, New York has never bothered to change the wording. It retains the usage both because of the inertia of the status quo and because the word gives leaders something to perpetually “fight” against—though clearly, they are not succeeding.

Yes, housing affordability is a problem—a huge ongoing problem—but the way its defined does nothing to solve it. If anything, defining the housing affordability problem as an “emergency” only serves to make affordability worse. It provides a convenient distraction to focus on the problem’s symptoms rather than its causes.

The Fire Extinguisher

To make an analogy, imagine a house with faulty wiring so that whenever a person turns on the lights, it produces sparks and ignites the wall. By having a fire extinguisher handy, one only needs to spray the dangerous spot when the switch is flipped. It’s thus easier to shout “emergency” and whip out the fire extinguisher—that is, to continually rely on rent regulations—than it is to address the fundamental problem—New York is not building enough housing, particularly in the middle-income areas.

The New York City Skyline. Photo by Jason Barr.

The Long Life of an Emergency

The history of rent regulations in New York can only be described by one word: labyrinthine. What began as a wartime expedient metastasized into a complex regulatory web, that evolved depending on the crisis du jour. In some years, rent regulations were particularly stringent; in other years, they would be relaxed a bit, only to have the exemptions rescinded when a new crisis arose. When housing costs spiked—usually because of economy-wide inflation—tenants would demand government intervention to protect them against rising costs.

Because the effects of inflation are sudden and noticeable and the benefits of new housing supply take years to materialize, officials realized that applying an addictive “pain killer” (or fire extinguisher) now was more politically saleable than the patience that generating new supply requires. Appeasing renters has been necessary because they comprise two out of three households. And yet, no matter how the rules changed, or how crises came and went, one thing stayed the same: rental vacancy has never risen above five percent since World War II.

1943 to 1968

New York’s decades-long history of rent regulations began in November 1943, when all rental units in New York had their rents frozen at their March 1943 levels as part of the U.S. Emergency Price Act of 1942. These controls were enacted because of the housing shortage and price spikes brought on by the war. Every effort was being used to fight Fascism, and normal, peacetime housing construction had to wait.

After the war, even as Federal rent rules expired, New York continued to implement rent controls for all units built before 1947, though new units were to be exempted. As housing expert Michael Stegman wrote in 1982:

The extension of rent control in New York City was intended to be a temporary measure justified by the continuing housing shortages related to emergency wartime conditions. Increases in supply through the normal workings of the market were expected to ease price pressures and allow the elimination of controls. “For this reason, a vacancy rate of 5 percent was established in the legislation as the criterion for the restoration of a normal supply-demand relationship in the housing market. Anything short of a 5 percent vacancy rate would be prima facie evidence of a continued ‘housing shortage.’”[1]

Yet, “temporary” rent controls continued for decades, though the rules on rent increases were tweaked over time. Typically, landlords could raise rents if a unit was vacated. At other times, landlords were allowed a one-time, across-the-board increase. Or raises could be implemented through negotiation with a tenant, usually based on an upgrade to the apartment. Tenants could also request reductions if they felt the landlord was not maintaining the unit. But in all cases, tenants in good standing had the right to remain in their apartments and could not be evicted. Units could even be passed from parents to children, essentially making rental apartments “co-owned” by landlord and tenant.

By 1965, 63% of Gotham’s rental stock was rent-controlled, 6% was in public housing, and another 5% were Single Room Occupancy (SRO) housing. Thus, only a quarter of the housing stock was unregulated.

1969 to the Present

1969 was a watershed year with the introduction of rent stabilization. All units completed between 1947 and 1969 were to be “stabilized,” and vacated rent-controlled units would also be stabilized. The rent stabilization program was an admission that rent control was a failure. It was clear by the late 1960s that the inability of landlords to raise rents according to their costs was causing the housing stock to decay. Now, all units in the program would have their rents increased by a fixed percentage each year set by the Rent Guidelines Board.

As a Rand Institute report wrote back in 1971:

The rental housing market in New York City is in a state of crisis. Vacancies are acutely scarce, construction is at its lowest level in many years, rents in the previously uncontrolled sector rose so rapidly in 1969 that a new form of control was imposed, and large numbers of recently habitable buildings have been reduced to shambles or withdrawn entirely from the market. Tenants are deeply dissatisfied either with the quantity of service provided by their landlords or with the rents demanded, or both. Landlords are equally dissatisfied with the yields of their property, the behavior of their tenants, the burdens of public regulation, and the illiquidity of their investments.

To help grease the housing market, the state passed a law in 1971 that decontrolled all apartments that were vacated as of June 1970 (with new construction exempt as well). However, in 1974, the law was changed once again: all units completed after 1969 were to be stabilized, and furthermore, newly constructed buildings would be stabilized if they received 421(a) tax abatements.

1997 to Today

Then, in 1997, the state allowed the decontrol of the more expensive units, and hearty rent increases for vacant units. But these rules was reversed in 2019, when the City and State doubled down on more stringent rent regulations. Since then, vacancy decontrol and decontrol of high-end units have been abolished. And now only a tiny fraction of the cost of major capital improvements can be passed on to tenants. The effect of these regulations has been to encourage the decay of lower-income apartment buildings. Today, over half of the housing stock is either in rent control (1%), rent stabilization (41%), or public housing (7.5%)—over a million units.

Most importantly, if the rent regulations law were designed to help the lowest income families, they were not doing their job. In 1975, for example, the majority of residents in the lowest income brackets were paying more than 40% of their income on rent. Today, the vacancy for those in the lowest rental brackets is 0.4% (basically zero). For renter households earning less than $70,000, more than half are paying 55% or more of their income on rent.

A Very Brief History of NYC Vacancy

There is nothing inherent in housing markets that requires rental vacancy rates to be systemically less than 5%. The long-run average vacancy rate for rental housing in the U.S. since 1956 is 7.3%. During this time, the lowest U.S. vacancy rate was 5%, and there’s no evidence that U.S. vacancy rates are declining. New York State’s average vacancy since 1986 has been 5.4%. In the 1910s and 1920s, Gotham’s vacancy rate was typically above 5%. Furthermore, New York City’s market-rate buildings frequently have vacancy rates above 5%.

Vacancy Rates Over Time. The blue line is the annual rental vacancy rates for the U.S. since 1956. The orange line is New York State rental vacancy rates since 1986. The green line is vacancy rates since 1996 for New York City unregulated units. The black dashed line is the benchmark rate of 5%. Sources: U.S. here, NYS: here, NYC: Compiled from various NYC Housing and Vacancy Survey Reports.

Period 1: The Roaring Twenties and the Great Depression

After the First World War, Gotham’s housing vacancy was near zero (see figure below). The war and dislocations caused by inflation, supply chain problems, and lack of credit resulted in the virtual cessation of new construction. But then came the building boom of the Roaring Twenties and by 1927, the vacancy rate was 5.16%, and in 1929 it was 7.7%. The high rates were driven primarily by the vast amounts of new housing built during the period.

But then came the Great Depression. People were out of work and were doubling up. By 1934, vacancy peaked at nearly 15%. It would continue to fall through the decade, and on the eve of World War II, it was a healthy 5.6%.

Period 2: Post-World War II to 1990

By 1945, however, New York City’s vacancy was near zero. The war effort drew people to the city while housing construction ceased. But new construction began to pick up after the war, and vacancies slowly started to move upward. Then came the Korean Conflict (1950-53); inflation rose quickly, housing construction plummeted, and rent controls were tightened. As a result, by 1955, vacancy was back down to almost zero and remained below 1.3% for the rest of the decade.

Then, changes in the zoning rules triggered a building boom. The 1961 zoning codes downzoned the city, and so developers rushed to get shovels in the ground before the new, stricter laws took effect. This produced a relative flood of new units, which caused vacancy to rise to 3.19%, a post-war high, though still not high enough to trigger the end of rent control.

After the construction boomlet, housing construction slowed, and vacancies fell and remained below 2% for the rest of the decade. In the 1970s and 1980s, vacancy remained below 3%, as housing abandonment skyrocketed, and new construction was muted.

Period 3: 1990 to the Present

Then came the New York Renaissance which started in the mid-1990s. While vacancies inched up from the historical 1970s lows—thanks to low inflation and increased construction—they have never reached 5%. They have showed a remarkable stability, hovering between 2.9% and 3.6% in the last thirty years. Even the apparent mass exodus caused by the COVID pandemic only brought vacancy rates to 4.5%. Recently, the renewed growth of New York, combined with the lack of new supply, along with the cessation of new construction tax abatements has brought vacancy down to 1.4%.

New York City Rental Vacancy Rates, 1909-2023. Sources: Data complied from several sources by the author (sources available upon request)..

A History of Housing Supply

So, what has kept vacancy below 5% for 80 years? Ultimately, it boils down to housing supply. In short, since World War II, New York has not been building housing at a fast enough clip. Let’s take a tour of Gotham’s supply since 1920. However, we will not look at the number of units built each year–which is what we typically hear about in reports about New York’s housing market–because it’s an incomplete aspect of the story.

A more accurate measure is the relative net supply—the number of new units minus demolished units as a proportion of the total number of existing housing units. In the figure below, I give two versions of relative supply: the total number of new (gross) units completed per 1,000 housing units and the total number of net units completed per 1,000, where net here is the total new units minus the number of demolished units.[2]

Net New Housing Construction Rates for New York City, 1920 – 2023. The graph gives the number of new units produced per 1000 units of existing stock (blue line) and net new units (new – demolitions) per 1000 units of existing stock (orange line).  Note: the black line is 20 units per 1000, a benchmark for net new construction that moves vacancy above 5%. Sources: Data complied from several sources by author (sources available upon request).

The Takeaways

The graph above provides a few key takeaways.

First, the last time that the number of net new units was above 20 per 1000 (or 2%) was 1931. In short, New York has been unable to build past that number in nearly a century. 20 per 1000 seems to be the “magic number” for getting vacancy above 5%.

Second, although there were two periods where gross construction between 1935 and 1965 hit the 2% threshold, new supply was in a big battle with units being demolished. Some of these lost units were from slum clearance projects (both Title I and public housing), but much of it was because apartment buildings were being abandoned. The demolition of older housing units basically stopped around 2000.[3]

In short, since World War II, there have only been two real supply booms, the one from around 1948 to 1952 and the second from around 1962 to 1965. But in neither case was the boom above the 20 per 1000 threshold.

Net Negative

Third, net housing construction was negative between 1977 to 1991. Thus, for a decade and a half, new supply was unable to match the number of units demolished, even as the population of New York was growing.

Since the mid-1990s, when New York’s economy and population started to rebound, net construction rates have hovered around 5 per 1000, about half as high as they were in the quarter century following World War II. The construction since 2000 has been positively timid.

While the recently passed City of Yes Housing Opportunity is a good start, it will not be enough. Under perfect conditions, it will raise annual construction rates from 5 per 1000 to 1 per 1000, still much lower than what’s needed to move the vacancy needle.

So, What Drives Supply and Vacancy?

In our review of New York’s housing market, we have seen how vacancy rates have remained below the “emergency” cutoff of 5% and have also seen that net housing supply has been consistently low over the decades. But the questions remain: What drive supply and vacancy increases, and how are the two linked?

In the next post, we will discuss the connection more thoroughly and see how New York can end its housing crisis once and for all.

 

—

[1] See Stegman (1982) for the source of the internal quote.

[2] Data on new units through subdivisions or building conversions or loss of units through combinations or conversions away from housing are not so easy to come by. However, based on my analysis, the gains and losses from these types of changes to housing tend to offset each other, on average.

[3] A frequent complaint in today’s market is that New York is losing moderately priced housing because of teardowns that are  replaced by luxury condos. However, looking at the data within a larger perspective shows that housing teardowns in the 21st century are at an all-time low. We should be less concerned about teardowns and more concerned about new construction citywide.

by Jason Barr 4 Comments

The Housing Twenties: New York’s Biggest Building Boom and Its Lessons for Today

Jason M. Barr                October 29, 2024

Let’s travel back a century—to New York during the Roaring Twenties.

On our journey, we will see a very different Gotham–not just flappers, speakeasys, and the new jazz–but more importantly massive housing construction. Arguably, 1921 to 1929 was the greatest housing boom in New York’s history. Units of all kinds—high-rises, garden apartments, and one- and two-family homes—were constructed in neighborhoods far and wide. It was a time of housing plenty.

On the other hand, today is a time of housing scarcity, with the housing vacancy rate at a mere 1.4% and with rents constantly rising.

The vital lesson from the Roaring Twenties is that supply matters. When new construction is robust and housing abundant, many benefits follow. First, rents fall. Second, there are more choices, so residents can find the housing size and location that best fits their needs—and they get more space for each rental dollar. Third, new construction upgrades housing quality, allowing households to live in safer, more user-friendly units.[1]

On the other hand, when housing is scarce, rents rise, and residents are forced to accept units that don’t match their needs. And they find it difficult to move when their life circumstances change because they have fewer options.

Where’s the Beef? Map of Housing Built in 1920s and since 2010. When compared to housing built during the Roaring Twenties, today’s housing is hard to find. Note the 1920s housing is extant housing, thus under-representing the true construction during the period. Source: NYC PLUTO file, 2024.

Housing Supply Then and Now

Let’s view the number of units completed in two respective 24-year periods, 1920-1943 and 2000-2023 (see the graph below). Several things jump out. First, the Roaring Twenties was roaring indeed. From 1920 to 1929, New York built over 740,000 housing units (when New York’s population was one-third less than today!). On the other hand, from 2000 to 2009—during the great housing “boom” of the early 2000s—New York built less than 200,000 units. (Data sources can be found here.)

In the 1920s, Gotham was building housing faster than its population growth. Today, the city barely builds, and population growth is a trickle. In the 1920s, when immigrants arrived, they found affordable housing and raised their children in the city. Today, housing is a game of musical chairs. People move in, bid up the price of scarce units, and others leave as a result. The only reason that New York’s population has grown in the 21st century is because of improved health—people are living longer; on the hand, net migration has consistently been negative. That is, year after year, more people leave then arrive.

The second fact that jumps out from the graph is that the period from the Great Depression to America’s entry into World War II (1930 to 1941)—arguably the 12 worst years in American economic history—surpassed housing construction in the best 12 years of the 21st century (2012 to 2023), with 279,134 units during the first period and 239,910 during the second. In short, New York’s 21st-century booms can’t compete with 20th-century depressions.

New Housing Units for Two Respective Periods. The Blue Line is from 1920 to 1943. The Orange Line is from 2000 to 2023. Thus 0 is both 1920 or 2000, respectively. Source: See here.

Vacancy

At the end of 1928, apartment rental vacancy was 7.8%, and nearly half the city’s vacant units were in the lowest-income rentals. Due to subway expansion and brisk housing construction, those with lower incomes could improve their housing situations relatively easily. From 1920 to 1929, Brooklyn alone built 250,000 units—more than all five boroughs from 2000 to 2009.

New York’s current low vacancy rate stems from the slowdown in construction due to the COVID pandemic, high interest rates, the rescinding of 421a tax abatements in 2022, and the renewed growth in the city’s economy. But the truth is that housing vacancy has been unhealthy for many decades. The Rent Stabilization Law defines a “housing emergency” when vacancy is less than 5%. By this definition, the city has continually been in a (self-imposed) housing emergency since World War II and has forgotten what a healthy housing market looks like.

But just as important, nearly all the city’s vacant units are at the high end of the spectrum. Vacancy rates for the lowest-income New Yorkers are way below 1%. With this rate, people have few, if any, opportunities to change or improve their housing.

Vacancy Then and Now. The left graph is housing vacancy vs. rent per room for 1928. The right graph is housing vacancy vs rent for 2023. The y-axes are aligned to show how scarce housing is today. The graph below is a larger version of today’s graph. Sources: Left: State Board Housing Report. Right: NYCHVS.

 

Vacancy Rates in 2023 vs. Asking Rent. Note that vacancy is above 1% only in the highest rent category. In other words, for low- and middle-income housing there is nearly no vacancy. Source: NYCHVS.

 

In With the New Out with the Old

As a bonus, new housing improves health and safety and creates a better match between what residents want and what’s available—tenants get new features and layouts that reflect their current needs and tastes. New housing is generally safer as it incorporates the most current building codes and health standards.[2] As the COVID pandemic has shown, many older buildings, with sealed windows and reliance on air conditioning, were not ideally suited for providing air circulation and outdoor space.

During the 1920s, low-income tenements were abandoned in droves. Officials relished the disappearance of the city’s worst housing. In a report written in 1928 by the New York State Board of Housing—not a group known for its support of tenement landlords—wrote:

Perhaps the most gratifying aspect of the improvement in available dwelling space is the increase in the number of vacancies in the old law tenements and the cheapest apartments generally. The prolonged period of prosperity, combined with the increased physical supply of housing, has made it possible to abandon some of the most unsuitable housing….In this connection, the significant fact is not that the poorer classes of the population now have greater freedom of movement and wider choice—for at best the choice of the lower rental offers very little—but rather that the economic position of at least a portion of the poor has been raised sufficiently to enable them to abandon so large a number of wholly inadequate apartments.

Scarce Housing Policy

On the other hand, today’s “scarce housing policy” causes low-income housing to deteriorate. The problem has been made worse by fiat. In 2019, the De Blasio administration enacted a law limiting landlords’ ability to pass along to tenants the cost of upgrades and improvements. As a result, landlords have been incentivized to worsen the housing stock. Tenants are now squeezed on two ends. They have little choice but to remain in their units because of low vacancy rates, and landlords have little choice but to undermaintain their buildings.

The most recent survey of New York’s housing stock shows a general deterioration in low-income housing in the last few years, which is no doubt tied to both the city’s low vacancy and the 2019 law. As the report states,

The 2023 [NYC Housing Vacancy Survey (NYCHVS)] showed a higher prevalence of most individual housing problems relative to earlier cycles, following the same general pattern as in 2021. This was true for housing types, including rent stabilized units and market rentals. The share of units reporting cracks or holes, broken plaster or peeling paint, and toilet breakdowns remained statistically similar to 2021.

Rents and Affordability

Arguably, the best measure of affordability is rent relative to income. Getting exact apples-to-apples data for the two periods is tricky, but we can generate relatively comparable measures. For the 1920s, I have compiled an index that tracks the average wages of industrial workers in New York to average apartment rents. For rents, I used two indexes. The first was created by the Bureau of Labor Statistics (BLS). The second is based on newly created data from four economists, Ronan Lyons, Allison Shertzer, Rowena Gray, and David Agorastos.

The BLS data tracks people who have been renting their units for some time. These people are more likely to experience fewer swings in their rent. During the bad years after WWI, rents rose less slowly (in part due to rent control laws and the fact that landlords were often reluctant to dramatically increase rents on longtime tenants). But during the good years, they reduced rents less quickly due to inertia.

On the other hand, the Lyons et al. index comes from asking rents in classified ads from historical newspapers, which reflect the most current market conditions. Spikes are likely to be greater in lean times, when inventory is scarce, and rent drops faster in flush times when housing is abundant.

Rising Rents is Not a Given

When we compare rent indexes to wages, we see that the 1920s was a period of affordability improvement.  In 1914, the index values were at 100; by 1928, both indexes were well below that. Viewing the housing scene from today shows the opposite pattern. Tracking median household incomes to median rents shows that they have consistently risen in the 21st century. Researchers and policymakers consider housing “affordable” when it can be provided at 30% or less of household income. Since 2000, the median rent-to-income ratio has been consistently above 30%. And rents relative income has dramatically risen to above 50% for the lowest-income New Yorkers.  The city’s scarce housing is, in effect, impoverishing a large swath of residents.

An Affordability Index. This graph shows an index of rents relative to average wages for industrial workers in New York from 1914 for 1928. As the building boom of the Roaring Twenties got underway, starting in 1922, affordability improved. The spikes after WWI reflect the economic difficulties driven by the war. Sources: See here.

 

Median Rents Relative o Median Household Income 1965-2023. Note above 30% is considered housing that is relatively unaffordable. Source: NYCHVS.

What’s the Difference?

Of course, the real estate market in the Roaring Twenties was different from today. First was that the zoning codes were relatively lax. Experts in the mid-20th century estimated that if the city were built out according to the zoning (and tenement housing) codes enacted in 1916, the city would hold somewhere between 55 and 77 million people!

In 1961, the city was strongly downzoned. By my estimate, the zoning codes today allow for a maximum population of about 12 million people, though this likely overstates the case because 60% of residential properties are either above their maximum allowable floor area limit or within 25% of it. These are properties that are illegal or economically unfeasible to densify.

Second, there was no rent control or rent stabilization law as we know them. There was a form of rent limits implemented in 1920 during the “famine” years following World War I, but they did not inhibit new construction when the timing was right. Today, landlords cannot legally evict tenants in rent-stabilized buildings, and who have no incentive (or ability) to move because there nowhere to go.

Third, the amount of land for development was much greater back then. By 1980 or so, most of the large pockets of residential undeveloped land had been filled in. In the Roaring Twenties, large swaths of land were still open for urban development, and developers were buying up these lots to erect large apartment buildings or entire neighborhoods. While Gotham still has a trove of underutilized land, converting it to more housing is difficult and expensive because of the inability to remove tenants from a building, the fragmented nature of lot ownership, and stringent zoning regulations.

Lessons for Today

The key lesson for today, however, is that when a city builds housing faster than its population particularly in the outer areas, prices drop or flatten, quality improves, and residents have more choices. If a city builds large quantities of middle-income housing, it frees up older units for lower-income residents (through filtering), and they get better housing than otherwise.

Unlike the Roaring Twenties, however, the New York faces the need to redevelop and densify in a built-up city. I have written widely on what the New York can do to fix this problem but let me itemize a few key steps.

Transit Oriented Development

First is to encourage transit-oriented development by the subway lines outside of Manhattan. Upzone these areas for mid-rise buildings.  92% of housing near subway stops outside of Manhattan is three stories or less. If each building was redeveloped by adding double the current stories (and units)—that is, from two to four, or three to six, the affordable housing crisis would be over.

Target Vacancy Rates

Second, city government needs to target neighborhood vacancy rates. One strategy is for the City to buy large, under-utilized lots and sell off the ground leases to developers for new middle-income housing. As the vacancy rates go above 5%, tenants will have more flexibility and choices, and developers will be freer to redevelop their properties to contain more housing.

Reduce Construction Costs

Third, the government also needs to reduce regulations that choke off the building of affordable housing; and it must offer construction subsidies to lower the high construction costs, with proportionally higher subsidies for buildings that will rent to those with moderate and lower incomes.

Plan for Growth

Lastly, the City must plan for growth. As neighborhoods add more residents, the government needs to be ready to invest in better transportation—like streetcar service, dedicated bus lanes, and more subway lines—and to add more schools, services, and parks to preserve the quality of life.

City of Yes?

To his credit, Mayor Eric Adams is promoting re-writing the zoning rules to allow for more housing across the city through his City of Yes Housing Opportunity program. However, even if fully passed by the city council, which seems doubtful, the program would only generate about 7,000 additional units per year–hardly enough to make a dent in the problem.

If New York really wants to make housing affordable it needs a new Roaring Twenties for the 21st century–there’s still time to get started!

 

 

—

[1] This is not to deny the charm of older buildings, but the key is for a city to find a better balance between preservation and accommodating the needs of the present.

[2] Having said this, building codes often create an undue burden for those constructing low-income housing. If regulations were reasonably pared back, low-income housing would be cheaper to build.

by Jason Barr 4 Comments

Does Upzoning Work?

Jason M. Barr                       August 14, 2023

Many cities around the world have severe housing affordability problems. One reason is overly restrictive zoning regulations. Zoning—the rules that dictate what can be built on each lot—was initially designed to control urban problems, including congestion, pollution, and shadows.

Over time, however, planners made common cause with property owners to tighten the rules. Planners did not like urban density and were all too happy to reduce it. Property owners realized that stricter codes could preserve their property values and exclude the poor and minorities from their neighborhoods.

By the middle of the 20th century, the most popular zoning rule was limits on the so-called Floor Area Ratio (FAR)—the permitted maximum floor area stated as a ratio to the lot size. For example, a maximum FAR of 1 means a one-story structure can cover the entire lot or a two-story structure can be built on half the lot. FAR caps and plot coverage maximums guarantee much lower population densities.

A Bridge too FAR?

The decades of downzonings—reductions in allowable FARs—after World War II, however, did not affect housing prices all that much since large industrial centers were emptying out, and the suburbs still had plenty of land on which to build. But then, in the 1990s, these places began to rebound. A recent wave of research, particularly in economics, has documented the unintended consequences of stringent zoning regulations, including higher housing prices, increased segregation, urban sprawl, and greater greenhouse gas emissions.

If the 20th century was the Age of Downzoning, then the 21st century is gearing up to be the Era of Upzoning. Cities worldwide have initiated loosening building regulations. Logically speaking, if downzoning has reduced construction, then upzoning should produce more housing. Or does it? Does simply turning on the spigot cause the water to flow?

Let there be Housing?

To answer this question is not so simple, however. One must be able to see clearly through the fog of urban change. Expensive cities are dynamic places: people constantly move in and out, and businesses grow and shrink. Housing affordability seems to worsen no matter what new policy or subsidy the government adds.

The typical observer sees the tall cranes erecting luxury condos while urban life gets more expensive. It must logically follow that those cranes are “causing” higher prices. However, demand and supply are in a constant wrestling match. When the demand side wins, prices rise, no matter how much new construction there might be.

The Scientific Method

To separate supply from demand and cause from effect requires some effort. Researchers must create a scientific setup—a treatment and control framework—especially concerning changes to the zoning rules. If particularly hot neighborhoods are targeted for upzonings, there needs to be a way to identify the demand (the hotness) from the upzoning impacts (more supply).

For example, when testing a new pill that aims to lower cholesterol, it’s best to compare a group of people with similar characteristics. The ideal experiment is to enroll in a study, say 100 men, who are very similar—in their 50s, overweight, and don’t regularly exercise. Half are randomly assigned the pill, while the rest are given a placebo. If the men who took the pill have lower cholesterol, on average, it strongly suggests that the pill works.

Zoning as Treatment

Recent studies on upzoning, discussed below, use this logic, although, unlike medical doctors, researchers can’t control the experiment to such a fine degree. Instead, they need a way to locate a reasonable control group similar to the “treated,” or upzoned, lots before the upzoning was enacted.

Several studies use the so-called border method. They compare blocks in the upzoned area to those just on the other side. Presumably, two blocks across the street are very similar otherwise. Another approach uses time as the difference. When areas are rezoned randomly at different times, one can compare neighborhoods before and after the rezoning. Finally, another method is to use a statistical algorithm to match upzoned parcels to similar non-upzoned parcels elsewhere in the city.[1]

With this in mind, we now turn to the findings.

Upzoning in Seattle (left) and rezoning in São Paulo (right). Source: Krimmel and Wang 2023) and Anagol et al. (2023). Note: “BAR” is Building Area Ratio (same as Floor Area Ratio).

Upzonings Around the World

New York

New York City, under Mayor Michael Bloomberg, from 2004 to 2013 rezoned nearly 40% of the city’s residential land. Central areas of Manhattan, Brooklyn, and western Queens, which had underutilized industrial sites, waterfront blocks, and transit-accessible neighborhoods, were upzoned by increasing the maximum allowable FAR for residential use.

Hsi-Ling Liao, in her 2023 paper, “The Effect of Rezoning on Local Housing Supply and Demand: Evidence from New York City,” compares upzoned parcels to those that were not upzoned but were within 1000 feet (305 m) of the boundary. She concludes that in the upzoned districts, “…the number of residential units increased by more than 4 percent seven years after upzoning. The parcels that are more intensively treated and receive a stronger boost in allowed residential capacity experience an 8 percent increase in housing supply.”

A very similar study was conducted by Xuequan Elsie Peng in her 2023 paper, “The Dynamics of Urban Development: Evidence from Land Use Reform in New York.”  She compares the new housing units in the upzoned blocks to those that are within 650 meters away (0.4 miles). Her findings echo those of Liao’s—in the upzoned areas after six years, the number of units increased by about 4%, while after 12 years, the number of units increased by about 8%, relative to the blocks that were not upzoned.

Peng also studies treatment intensities by investigating if more generous FAR allowances lead to more housing. She finds that in the low upzoning case (FAR increases of less than 30%), there was about 2% more housing, on average, after ten years relative to the non-upzoned blocks. In the high upzoning case (FAR increases greater than 92%), she finds that these neighborhoods added about 8% more housing, on average, after ten years than non-treated blocks.

Portland

Hongwei Dong, in his 2021 paper, “Exploring the Impacts of Zoning and Upzoning on Housing Development: A Quasi-experimental Analysis at the Parcel Level,” studies the case of Portland. He focuses on parcels that were upzoned between 2001 and 2002 in southwestern Portland.  In this case, the rezonings were primarily in suburban neighborhoods, where a majority of lots contained low-density, single-family homes.

After matching upzoned parcels to similar non-upzoned ones, Dong concludes that upzoned parcels were about two times more likely to be developed fifteen years after upzoning and that “further analysis suggests that the 111 upzoned parcels produced 240 dwelling on 33.8 acres of land. The density is 7.1 units per acre. The fifty-eight control parcels produced eight dwelling units on 18.6 acres of land. The density is 4.3 units per acre. These findings suggest that upzoning not only sped up housing developments but also increased housing production.”

Seattle

Jacob Krimmel and Betty Wang study the case of rezonings in Seattle in their 2023 paper, “Upzoning with Strings Attached: Evidence from Seattle’s Affordable Housing Mandate.” Unlike New York and Portland, Seattle took a different approach. In 2019, select neighborhoods were upzoned to allow more floor area, but this benefit came with a “tax”—a mandatory housing affordability (MHA) requirement for developers to either include affordable units or pay into a fund for the city to spend on affordable units elsewhere. The authors write, “The reform combines two policy levers that some economists consider to conflict with one another: Increasing development capacity through upzoning and requiring private development to create income-restricted affordable housing.”

The MHA areas were applied to areas already zoned for multifamily, commercial, or high-density single-family homes (such as townhouses). However, the upzoning was relatively mild, amounting to one additional floor than otherwise, on average. Comparing blocks that were in MHA areas to those that had no change, they conclude that the program, in fact, backfired in that it drove new construction away from the MHA areas. They find

strong evidence of developers strategically siting projects away from MHA-zoned plots – despite their upzoning – and instead to nearby blocks and parcels not subject to the program’s affordability requirements. The differential reduction from MHA to non-MHA zones could be as large as 70% of average permitting activity at the border.

They further find that:

Worryingly, most of the drop in the number of units in MHA zones is coming from the multifamily segment of the market, where most of the housing products are 3- and 4-story townhouses and duplexes. This is of particular note because lowrise and small multifamily homes are seen as a more affordable alternative to luxury apartments for low- and moderate-income renters.

Chicago

In “Upzoning Chicago: Impacts of a Zoning Reform on Property Values and Housing Construction” (2019), Yonah Freemark explores the case of Chicago. Under Mayor Rahm Emanuel, Chicago undertook two rounds of upzonings. The first, in 2013, upzoned neighborhoods between 600 feet (183 m) and 1,200 feet (366 m) of metro and commuter rail line stops to encourage transit-oriented development (TOD). The second round, in 2015, expanded the upzoned areas to a half mile within the stops.

The rezonings involved two types of changes. The first was an increase in the allowable FAR, which in 2013 amounted to 17%, on average, while the 2015 rules increased the FAR from before 2013 to 33%, on average. Additionally, in both years, rules on on-site parking were made less stringent, allowing developers to provide fewer spots.

Freemark first compares parcels upzoned in 2013 to those left unzoned but were later upzoned in 2015. He then compares the parcels upzoned in 2015 to neighboring blocks just outside. For the 2013 upzonings, Freemark does not find evidence that they led to an increase in housing supply, as measured by new building permits.

However, this finding is likely due to three factors. First, the upzonings applied to both high- and low-income areas. Upzoning low-income areas is less likely to be effective if the income there is not high enough to compensate for the cost of new construction. Second, the period of investigation was from late-2013 to late-2015, a very short period to analyze the impacts of upzoning. Lastly, the average upzoning of 17% was rather small to have a meaningful impact, as other studies have shown.

However, in his analysis of the impact of the more generous 2015 upzonings, he finds (weak) evidence that the combined relaxation of parking and FAR rules increased the number of residential building permits in the nearly-three-year period after the rules changed (about 25 more permits in the upzoned areas versus the non-upzoned areas between September 2015 and June 2018).

Zurich

In their 2023 paper, “Making Housing Affordable? The Local Effects of Relaxing Land-Use Regulation,” Simon Büchler and Elena Lutz investigate the Canton of Zurich in Switzerland from 1995 to 2020. The canton contains 168 municipalities with a population of 1.52 million. Each municipality separately undertakes rezoning approximately every 15 years.

The authors exploit the independent and random timing of the rezonings to explore how increases in the FAR before and after the rezonings impact housing construction in 100-meter (328 feet) by 100-meter grid cells. For each cell, they measure the total floor area in five-year intervals and then track when the upzonings occurred. Thus, the treatment is the cell after upzoning, and the control is the cell before upzoning.

Their conclusion: “We find that large upzonings of more than 50% significantly increase the living space and housing units by 9-18% on upzoned parcels compared to non-treated parcels in the subsequent six to 15 years….In contrast, small upzonings, of 0-50%, do not incite a strong reaction from developers.”

Auckland

In their 2023 paper, “The Impact of Upzoning on Housing Construction in Auckland,” Ryan Greenaway-McGrevy and Peter Phillips study upzoning in Auckland, the largest city in New Zealand, with about 1.57 million residents. In 2016, the metropolis upzoned nearly three-quarters of its residential land.

The authors’ approach is to look at new building permits inside the upzoned areas versus those just outside. Studying the relative changes in annual building permits issued for new dwelling units, they conclude,

The response to the zoning change is immediate: The estimated treatment effects increase every year after implementation. By 2021, five years after the policy was introduced, some 23.06 additional permits are issued, on average, in upzoned areas compared to non-upzoned areas in each of the 479 Sas [Statistical areas, similar to census tracts]. This would correspond to 11,044 permits across the city. Cumulating the corresponding figures for 2016 through 2021 yields 34,614 additional permits in upzoned areas compared to non-upzoned areas. Treatment effects for attached dwellings exceed detached from 2019 onwards. By 2021, the estimated treatment effect for attached is 17.96 permits, while for detached it is 5.09.

São Paulo

Finally, in their 2023 paper, “Estimating the Economic Value of Zoning Reform,” Santosh Anagol, Fernando Ferreira, and Jonah Rexer study the case of São Paulo, the world’s fourth largest city, with 21 million residents in its metro area. The city undertook a rezoning in 2016. Similar to the other studies, they explore how raising the Floor Area Ratio—what they call the Built Area Ratio (BAR)—impacted new construction. On average, they found that the upzoning increased the allowable BAR by 36%, and more than half the city’s blocks received an increase.

The authors explore new housing permitting in upzoned blocks compared to non-upzoned blocks on the other side of the border. Their conclusion:

We find that the 1.4-point max BAR increase at the cut-off causes an increase of .003 multi-family dwelling permits filed per block per quarter, which is a sixty six percent increase in permits per unit of max BAR in treated relative to nearby control blocks. The differences in permitting activity between treatment and control blocks emerge just one year after the zoning reform was passed and increase thereafter, which is plausible given the time it likely takes for developers to create project plans, acquire land, and so on.

Upzoning and Housing Prices

Several studies also look at the impact of upzonings on real estate prices. Here the findings are mixed. However, interpreting price impacts is more complex. When rezonings occur, several processes are at work. The first is that upzoned parcel has the future opportunity—the so-called option value—to be redeveloped more densely and generate more revenue. A forward-looking market naturally means that upzoned parcels will sell at a premium, even if there is no redevelopment in the short run.

Second, upzoned neighborhoods typically mean more density, which means more Starbucks and other urban amenities, which raises local housing prices. There’s also the possibility of the opposite effect, where density causes more traffic congestion, street noise, crime, or trash—a so-called negative amenity effect—which will reduce prices (however, I sense is this is less of an issue today).

Then there is the supply effect: more housing will reduce prices, all else equal. However, when upzonings are small and take a long time to yield fruit, the supply effect may be swamped by the option value and amenity benefits.

The Findings

Liao’s study of New York City looks at the sale prices of buildings built before 2004 and thus excludes any buildings completed after the rezonings. She finds no statistically significant changes in housing prices in the upzoned areas, suggesting that the supply effect balanced out the amenity and option value effect.

Freemark’s study of Chicago found that after the upzonings in 2013,  density bonuses increased prices by about 16%, on average, vis a vis the non-upzoned parcels. But focusing on location, he finds that the result was likely driven by price rises only in Downtown, with no increases in other neighborhoods. And zooming in only on condo sales, he finds that prices increased by 12%, on average. Since the period of his analysis was short, it suggests that Downtown condo buyers had higher expectations for their neighborhood that showed up quickly in the sale prices but were not offset by more supply.

Büchler and Lutz’s study of Zurich looks at average rental listing prices. They found that rental prices in upzoned cells typically saw no price changes over time. Again, the supply effects of lower rents were evidently matched by the amenities benefit.

Anagol et al. look at neighborhood housing prices for São Paulo and find that areas with larger BAR increases had greater increases in the number of home sale listings, where a 1-point increase in the maximum BAR was associated with a 10.9% increase in listings growth, on average. And just as importantly, with more listings and supply, they found that each one-point increase in maximum allowable BAR was associated with a 5.7 percentage point reduction in home prices.

All told, these studies suggest that price impacts hang in the balance between the different supply and demand forces, but they also demonstrate that price increases are not a foregone conclusion.

Upzoning Effects: Each graph shows the rise in housing supply after upzonings (on the right half of each graph). Upper left: Büchler and Lutz (2023) for Zurich, middle left: Peng (2023) for NYC, upper right: Liao (2023) for NYC, bottom right: Anagol et al. (2023) for São Paulo, bottom left from Greenaway-McGrevy and Phillips (2023) for Auckland.

Does Upzoning Work?

This research shows that, yes, upzoning works. Of the seven studies reviewed here (excluding the Krimmel and Wang study), all but one found a statistically significant uptick in new housing permits, completed units, or residential floor areas after upzonings—and because of the upzonings.

But, of course, there are nuances and caveats. The studies reveal that the units materialize only after several years, and their impacts are relatively modest. But they do show that if strategically managed, upzonings can be effective.

What are the key lessons from these studies?

Scale

The first is that scale matters—the greater the allowable floor area increases, the greater the impact the upzoning will have. For example, Liao and Peng found that small FAR increases yielded much less housing for New York than larger FAR increases. Büchler and Lutz found no effect in the Canton of Zurich for upzoning of less than 50% but saw about a 10% increase after ten years for large upzonings. In Chicago, upzonings of 33% seemed more effective than of 17%. And Krimmel and Wang show that mild upzonings with additional costs can backfire by diverting construction away from these areas.

Scope and Location

While these studies do not directly investigate the amount of total upzoning versus total new supply, the evidence suggests that broader is better because of housing supply’s connection with demand. In central neighborhoods, a new high-rise can trigger an upward price spiral because of the national or international demand.

Outside of the center, where demand is more tempered, and land values are lower, new supply will be more easily felt because the new tenants are less likely to trigger an upward price cycle. On the other hand, in locations where the cost of new construction is too high relative to the expected rents or incomes, upzonings may have little effect. Districts with high owner-occupied and single-family homes will likely also see slower impacts because redevelopment can only occur once owners decide to move, which can take several years.

These studies focus only on upzonings. But many cities downzoned some neighborhoods simultaneously with their upzonings. As a result, rezonings can wind up shifting supply rather than boosting it citywide. And if rezonings push construction from the periphery to the center, price relief may not result.

An Affordable Future

Upzoning is necessary for a healthy city to grow and accommodate more residents, but it is not sufficient to solve the affordability problem. Older, denser cities still have a host of other barriers or frictions that make new construction difficult and slow, particularly in neighborhoods where it’s needed most.

Parcels must be assembled, buildings must be emptied, and other building regulations must still be followed—producing a steep mountain to climb.

Greasing the Wheels

In other blog posts, I have discussed steps that cities can take, but let me highlight a few that can help grease the wheels:

  • Link upzonings to neighborhood housing vacancy rates. The lower the vacancy rates, the more generous the increases in the FAR. This will ensure that upzoning is applied where it’s needed most.
  • Construction subsidies should be the greatest in the middle- and moderate-income neighborhoods. Abundant affordable housing comes from the construction of middle-income housing outside the center. Lower-income residents will gain through the so-called filtering process, where the older housing becomes more affordable, and people move into the new middle-income housing.
  • Homeowners should have much more flexibility to add Accessory Dwelling Units (ADUs), such as converting a garage to an apartment, adding a “granny flat” in the yard, or subdividing their house to add an apartment. These new units, which don’t require demolitions or evictions, are a relatively easy way to provide affordable units and generate extra income for homeowners in the suburbs.
  • A land value insurance program can insure homeowners against reduced land values if rezonings should cause neighborhoods to become less valuable. This will reduce (though not eliminate) NIMBYism since residents have some economic protection against change.

Leadership

The real problem of housing affordability, however, is not economics, it’s politics. Upzonings are contentious because people see them as generating higher housing prices and bringing changes to their neighborhoods. As a result, they elect leaders who either enact NIMBY policies or merely tiptoe through the garden of affordability.

True affordable housing comes from expansive citywide policies. The only way to do this is for leaders to form coalitions, seek compromise, create carrot-and-stick programs, and compensate those harmed to prevent them from stopping reforms.

Our cities cannot afford the delay.

 

Works Discussed

Anagol, S., Ferreira, F. V., and Rexer, J. M. (2023). “Estimating the Economic Value of Zoning Reform.” National Bureau of Economic Research, Working Paper No. w29440.

Büchler, S., and Lutz, E. C. (2023). “The Local Effects of Relaxing Land Use Regulation on Housing Supply and Rents.” Working Paper (July).

Dong, H. (2021). “Exploring the Impacts of Zoning and Upzoning on Housing Development: A Quasi-experimental Analysis at the Parcel Level. Journal of Planning Education and Research.

Freemark, Y. (2020). “Upzoning Chicago: Impacts of a Zoning Reform on Property Values and Housing Construction.” Urban Affairs Review, 56(3), 758-789.

Greenaway-McGrevy, R., and Phillips, P. C. (2023). “The Impact of Zoning on Housing Construction in Auckland.” Journal of Urban Economics, 136, 103555.

Krimmel, J. and Wang, B. X. (2023). “Upzoning With Strings Attached: Evidence from Seattle’s Affordable Housing Mandate.” Working Paper (July) and summary in Cityscape (2023).

Liao, H-L (2023). “The Effect of Rezoning on Local Housing Supply and Demand: Evidence from New York City.” Working Paper (July).

Peng, X. E. (2023). “The Dynamics of Urban Development: Evidence from Land Use Reform in New York.” Job Market Paper (January).

 

—

[1] The matching method used by Dong is called propensity score matching. The rest of the papers used what is called Difference-in-Difference (DiD) regressions.

by Jason Barr 6 Comments

The Missing Bottom: Filtering and Housing Affordability

Jason M. Barr                   April 3, 2023

“More Rundown Buildings Now!”

That should be the new motto for expensive cities. In the current debates on affordable housing, there is no discussion that the best way to help those with the lowest incomes is by letting some of the housing stock decay to a lower quality. The process of new housing for upper and middle-class households eventually becoming low-income housing is called filtering. For many decades, filtering was the primary route to affordability.

Current policies, such as mandatory inclusionary housing and tax breaks for new low-income units, don’t work because they require massive resources to help only a handful of families. The fact that cities hold lotteries for the precious newly built units is prima facie evidence of the absurdity of using new housing to solve the affordability problem. Rationing gems to the lucky few illustrates that the housing market is failing everyone else who must duke it out for what remains.

Decay Can be a Force for Good

The idea that we would let some of the housing stock decay to provide affordable homes was abandoned decades ago. Filtering was, after all, associated with slums, and slums had to be cleared. But to make housing more affordable for everyone, we must think of housing not as an infinitely lived product but as one with a life cycle requiring planned obsolescence. Cities need to expect and create the conditions to allow their housing stocks to cycle over, say, 50-year periods—to let new housing wear down and worn-out housing be replaced.[i]

Thus, filling in the “missing middle” in the housing market is doubly essential. It will not only provide new units for the middle class but will also generate a pipeline of properties that will filter down to the rest.

As I will discuss below, filtering is still vital for millions of residents across America. It’s just not allowed in the places where it’s needed most—in big, booming cities like New York. It’s time to make filtering cool again.

The Lower East Side circa 1900. This Manhattan neighborhoods was arguably the world’s most crowded, yet it was able to provide affordable housing for  immigrant communities because of the filtering process.

The History of Filtering: Manhattan 1900

In 1900, Manhattan’s immigrant neighborhoods were arguably the densest places on Planet Earth. Block after block was filled with tenements, each chockablock with residents living in tiny, dark units.

Jacob Riis’ 1890 photo essay in his classic book, How the Other Have Lives, advertised the seedier side of tenement life and helped shape public opinion that the slums had to be cleared. He was one of a group of reformers who were convinced they could save the city by surgically removing worn-down, low-income housing. Their goal was to scatter the population into new suburban homes and transform the metropolis into a worker’s paradise.

Yet, despite the overcrowding and the constant influx of immigrants, the truth was that the market was providing affordable housing. One 1907 study of 200 households living in the tenements found that the average family with six people paid 20% of their family income on rent. Only the poorest of the poor paid more than 30%; even in the worst case, it was never more than 43%.

This contrasts with New York City today, where 54% of households are rent burdened in that they pay more than 30% of their income toward rent, and 28% are severely rent burdened in that they pay more than 50% of their income on rent.

The housing was not pretty, to be sure, but it was available and relatively cheap. And just as importantly, the small apartments were considered temporary. The expectation was that families would rent up or move out as their incomes rose. As one investigator of immigrant neighborhoods opined in 1901,

Many tenements in Jewish quarters are owned by persons who formerly lived in crowded corners of others just like them; and from this population comes many and many a Broadway merchant and professional men in plenty. It is a common saying that from Hester street to Lexington avenue is a journey of about 10 years for any given family.

This was only possible if housing was cheap all around.

Supply in the Slums

Several things made sure that affordable housing was available. First was the continual process of generating new supply. And second, was that the buildings were disposable. They were constructed, allowed to depreciate, and, at some point, torn down and replaced with new housing.

To reformers, this idea was anathema, as they were laser-focused on housing quality and the problems of disease and vice. After an investigation of the tenements in 1903, one reformer, Elgin R. L. Gould, blamed the landlords for not taking care of their properties:

The building being flimsily built and the class of people living in it being somewhat destructive, when subject only to the poor management which ordinarily prevails, the item of repairs at the end of the year is large. Removals are frequent, as a slight inducement offered by a rival owner will cause a tenant to pack up and go, particularly to a new building just completed.

But the landlords were just supplying what the immigrants could afford. The constant churning meant that the housing market could fluidly adjust to cater to the evolving needs of the immigrants who poured into the neighborhoods.

Lodgers

Many households took in lodgers. One study of New York working-class families in 1909 showed that 34% of households took them in. Ironically, those most likely to have lodgers had relatively higher wages, as it seems the higher income was used to rent a place slightly larger than needed to let out the rest for a profit.

This provided more income for them and offered the smallest and cheapest housing possible for those who had just arrived from the Old Country. Lodgers represent the ultimate matching of supply and demand with those who were just getting started in America.

The Tenement Lifecycle

Tenement housing was expected to depreciate. There was no way to stop it since, as Gould documented, the residents had low incomes and moved relatively frequently, generating significant wear and tear on the structures. As such, it was typical for an entrepreneurial immigrant to buy an old apartment building, tear it down and build a new one. Relatively well-off residents would move in and pay the higher rent needed to cover the construction costs.

Over the following decade, the building would slowly depreciate, and the rental income would fall. But by then, the mortgage was paid off, and the payback period for the initial investment was returned. From there on, the property was “milked.” As long as the rental income—however low it might be—was greater than the operating costs and the opportunity cost of holding the land with that building, the owner could provide poor-quality but cheap housing.

At some point, the milking process had to end. When the building was in such bad shape, it would be torn down, and the process would start over. And in this way, the Lower East Side tenements had various housing qualities to satisfy the range of demands.

And just as importantly, even during the period of the greatest immigration, when new households were absorbing thousands of units, the market still was affordable because it was flexible. Not pretty, but flexible. And it shows that even in markets with tremendous demand, a city can build to accommodate everyone if it builds rapidly and expansively.

Reformers and Filtering

In the early 20th century, housing reform began to remove filtering as a useful housing option in big cities. Filtering never disappeared, but it slowed down so much that when cities like New York began to rebound in the 1990s, there was very little low-income housing left. What remains is “rationed” through rent controls.

A lot of good came from reform, of course. Laws mandating minimum sunlight, toilets, and hot running water in every apartment were not only done for humane reasons but also to improve community health. But stringent zoning and building regulations, along with slum clearance programs, were designed to prioritize quality over quantity—laudable goals, but not without unintended consequences.

The lasting implications of these policies were that they created limits on new housing construction, particularly in low-income neighborhoods, and they limited the flexibility of building owners to subdivide or adjust their properties based on the needs for different kinds of housing. The result has been to remove affordable housing from the total stock.

Filtering Across America. This graph presents estimates of average filtering rates for rental units based on the incomes of in-movers. On average, the income level for a household moving into a 50-year-old home is 30% that of a household moving into a new unit. Source: Rosenthal (2014).

Filtering in the 21st Century

In recent years, a handful of economists using cutting-edge statistical methods provide clear evidence that filtering not only still exists but also works to provide low-income housing when allowed to do so.

Since it is nearly impossible to get data on the specific nature of housing quality over time, an important 2014 study by the economist Stuart Rosenthal looked at the (inflation-adjusted) income of people who buy or rent housing over time within a sample of the same units. His data set contains over 72,000 housing units in a nationally representative sample between 1985 and 2011, and he then obtains the income of those who moved into these units over time.

His data show that, across America, the average change in moving-in household income is -11.8% for rental units and -7.5% for owner-occupied units. That is, on average, those who moved into these same units had lower income each subsequent round in large part because the housing was becoming older and more worn down.

And using more sophisticated statistical (regression) methods, he finds that filtering is most rapid in the first 50 years of a home’s life. After that, the filtering process typically stops and then reverses itself as homes become prized for their historical appearances or demonstrate that je ne sais quoi that makes them worth preserving or upgrading.

Furthermore, he finds that annual filtering rates are much faster for rental versus owner-occupied units—about roughly -2.5% per year versus 0.5%, respectively.  He concludes that across the U.S., on average, “[t]he real income of a newly arrived occupant at a 50-year-old rental unit would be just 30 percent of arriving occupant income in a newly built home….”

Filtering 2.0

In a paper published in 2022, three economists studied the issue of filtering in the same vein as Rosenthal. However, their data set was much larger—some 1.23 million household “turnovers” in their sample of 1.1 million housing units. They look only at owner-occupied units, but the findings are clear. On average, the rate of filtering for the U.S. housing stock is about -0.42% per year, similar to Rosenthal’s finding.

Within Cities

Because the authors obtained such a large sample, they estimate filtering rates across metropolitan areas and within cities. They find that filtering rates vary widely. Some neighborhoods see rapid filtering rates, whereas others see an upgrading of their housing stocks (call it reverse filtering).  In fact, they find that the variation in filtering rates is greater within cities than across them.

Though they do not do a deep dive into what drives this variation, one key finding is that central city neighborhoods have seen significant reverse filtering. This makes sense. As the desire to be in central cities has risen, it has driven up land values and housing prices.  Thus, only those with relatively higher incomes can afford to move in. More broadly, filtering is virtually impossible in neighborhoods where housing prices are rapidly increasing.

Across Cities

But just as interestingly, they explore the reasons why metropolitan areas across the country have very different average filtering rates. The main finding is that the ability of a city’s housing stock to keep up with demand—its so-called supply elasticity—is a vital determinant of its filtering rate.

One of the most important factors is the restrictiveness of zoning and the stringency of building regulations. Places with higher stringency have higher prices and less filtering. Thus, in metropolises like New York, San Francisco, and Seattle, where there is relatively little new construction, filtering is greatly diminished and reverses itself, on average.

 

Filtering and its Correlates. These graphs show that there is less filtering in cities that have rapid population and housing price growth, as shown in (a) and (d), respectively. Cities with difficulty is adding new housing and with strict building regulations have less filtering, as shown in (b) and (c), respectively. Note that positive numbers on the y-axes denote reverse filtering–that is, owners are upgrading their housing or that higher-income people are moving into older houses. Source: Liu, McManus, and Yannopoulos (2022).

Moving Chains

To get more evidence on filtering, we can turn to two recent studies that look at so-called moving chains. The idea is to follow the people, not the houses. When a new high-rise apartment building is opened, it initiates a moving chain, where a large fraction of units is occupied by locals who move from other parts of the city. When this happens, the “abandoned” housing is freed up and occupied by someone else, who then frees up their housing for someone else, and so on.

Two new studies statistically document these moving chains and look at the income levels of the people who occupy each vacant unit in the chain. They find that when a new high-price, centrally-located building is created, people of lower income are able to benefit from the empty units in the chain.

Helsinki

One study focuses on the Helsinki metropolitan area (HMA) with about 1.2 million residents. They begin by looking at 106 newly construed multi-family buildings, with 3,196 new units, built between 2010 and 2019 within a 3-kilometer (1.8-mile) radius of the Helsinki Central Station.

As expected, residents in the upper-income brackets tend to occupy these new units. However, they find that

the moving chains triggered by these new units reach middle- and low-income neighborhoods. By round three, 60% of movers originate from neighborhoods in the bottom half of the neighborhood income distribution….By round four, 50% of movers are ranked in the bottom half of the national level household income distribution.

In plain English, the new high-priced units allowed filtering for lower-income people.

The U.S.

Well, maybe, you say, there is something special about Helsinki. Another recent study looks at the moving chains across the U.S. In this case, the economist Evan Mast followed the “hopping” created by the opening of one of 686 market-rate multi-family constructed between 2009 and 2017 in the center of 12 large U.S. cities.[ii]

He finds that 67% of the new units in the central city are occupied by people already living in the metropolitan area. Even in global cities, the proportion is high. In New York, for example, 76% of new-building occupants are from within the New York metro area, and 68% are from within the city itself.

Following people along the moving chains, Mast finds that,

Constructing a new market-rate building that houses 100 people ultimately leads 45 to 70 people to move out of below-median income neighborhoods, with most of the effect occurring within three years. These results suggest that the migration ripple effects of new housing will affect a wide spectrum of neighborhoods and loosen the low-income housing market.

Again, in plain English: New upper-class housing in central cities benefits those in low-income neighborhoods through the filtering process.

We can’t see it amidst the fog of our everyday urban experiences, but if you telescope on filtering, it’s there. It’s just not happening fast enough to keep up with the demand for new housing.

Policy Implications

Research and history suggest that we need to think about housing and cities not as static entities but as dynamic systems. Newly constructed middle-income housing not only benefits those in the middle class but is also the primary means by which low-income housing gets produced. As the research on moving chains shows, we should not fear high-rise luxury construction in gentrifying neighborhoods, though we cannot rely on it either.

Middle-class housing—and lots of it—needs to be built where land values make it affordable to construct, primarily outside the central areas in urban residential and suburban neighborhoods and ideally along key transportation routes and mass transit lines.

The key is to establish a housing market that can quickly and flexibly adapt based on demand conditions. To this end, here’s what cities and states can do.

Flexible Housing

  1. Allow—even encourage—the construction of accessory dwelling units (ADUs) and the subdivision of larger homes into two- or three-family houses. These extra units can generate additional income for the property owner and provide a small and cheap unit for people who need it.

 

  1. Reform zoning to allow and promote the construction of mid-rise apartment buildings around transportation hubs and along transportation lines, particularly in the suburbs. Cities need to relax their overly stringent zoning regulations massively. Full stop.

 

  1. Allow for office buildings to contain mixed uses. All those half-empty office buildings should be allowed to have several different uses. Keep the offices on the lowest floors and maybe even at the very top ones too. Allow for shopping, schools, and retail in the middle, and permit housing everywhere else. If landlords can earn money by allocating floors to different uses, they should be allowed to do so.

 

  1. Encourage the construction of rental units rather than owner-occupied ones since middle-class rentals are more likely to filter down to low-income housing. Ways to do this include providing more generous subsidies to developers of middle-income rental housing and establishing tax-free renter savings accounts for tenants, where an extra part of their rent payment is set aside each month in a savings account (and can be used to pay rents in a time of emergency).

 

  1. Move to a land value tax program that taxes the land values at a greater rate than the building value. This taxing scheme will incentivize property owners to densify their land, providing more housing.

Flexible Cities

But as the fierce debates between the Nimbyists and Yimbyists show, change does not come easily. Making the housing market more flexible requires people to be confident that changes will improve their lives. To this end, governments need to create policies that enhance trust and mitigate fear.

As I have discussed elsewhere, two key elements include creating a new type of master plan and a land value insurance program. The master plan would, among other things, provide a blueprint for how new schools, parks, and local services get produced as population density rises. The point is that cities need to get ahead of the services curve to show that quality of life can improve even as neighborhoods densify.

Additionally, cities need to allow property owners to buy land-value insurance, protecting them against possible, though unlikely, drops in their land value because of neighborhood changes. Since homeowners fear drops in their house values, they prevent more housing in their neighborhoods. If they had insurance, they would be more likely to go along with rezonings, densification, and allowing the filtering process to work.

Affordable Cities

If history is any guide, big cities like New York need not suffer from affordability problems. But fear of change and stringent regulations have gummed up the works. Leaders need to do a better job in demonstrating that the lack of affordable housing is harming everyone, not just those in the missing bottom.

New policies cannot focus only on housing but must also address the fear of change and the additional services that will be needed as neighborhoods densify. After all, cities are not just places with houses, but are places where people live.

While filtering has gained a bad reputation because of the historical problems of overcrowding, it has traditionally been the primary source of low-income housing. We need to re-learn this if we truly want our cities to be more affordable.

  

Works Cited

Bratu, C., Harjunen, O., & Saarimaa, T. (2023). “JUE insight: City-wide effects of new housing supply: Evidence from moving chains.” Journal of Urban Economics, 103528.

Liu, L., McManus, D., & Yannopoulos, E. (2022). “Geographic and temporal variation in housing filtering rates.” Regional Science and Urban Economics, 93, 103758.

Mast, E. (2021). “JUE Insight: The effect of new market-rate housing construction on the low-income housing market.” Journal of Urban Economics, 103383.

Rosenthal, S. S. (2014). “Are private markets and filtering a viable source of low-income housing? Estimates from a ‘repeat income’ model.” American Economic Review, 104(2), 687-706.

 

——

[i] There are carbon implications to this. On the one hand, as old housing is destroyed or completely refurbished, it can be upgraded to the latest standards that minimize energy usage. Also, except for very tall buildings, mass timber structures can be used to minimize embodied carbon. For that matter, densification is good for reducing carbon footprints.

[ii] These cities are New York City, Chicago, Dallas, Houston, Washington, Philadelphia, Atlanta, Boston, San Francisco/Oakland, Denver, Seattle, and Minneapolis.

Buy the Book

Building the Skyline can be purchased through Amazon or Oxford University Press.

Credits

The New Amsterdam map on this page was sourced here, and the atlas image is from the 1891 Atlas of City of New York by GW Bromley & Co.

Website by
SQUAREWATER

Social media & sharing icons powered by UltimatelySocial