Tackling Homelessness in all its Complexity

A record number of people nationally are experiencing homelessness. In New York City, the Department of Homeless Services (DHS) shelter census stands at approximately 83,000, which is below recent peaks but still extremely high.¹ Homelessness is immensely costly – to the families and individuals experiencing trauma and dislocation, to children who manifest adverse impacts of early childhood homelessness over time, to all levels of government using public resources to attempt to manage the problem, and to communities wrestling with people living in spaces not meant for habitation.²  

The federal government, states, and communities across the country have all worked on plans to end homelessness, yet the number of people experiencing homelessness is increasing.³ Limited dedicated resources and political constraints are of course part of the reason for this lack of progress but do not tell the full story.  

After three years leading the nation’s largest homeless services agency, I believe we need to do a better job recognizing the complexity of homelessness in order to make progress. Specific policies and programs will vary based on local context and resource availability. But the themes explored below are cross-cutting and should serve as a foundation for local decision-making that takes into account homelessness in all its complexity.  

1. The at-risk universe is far larger than we typically discuss.

Measures of housing instability and risk for homelessness generally focus on rent burden, at least in part because the number of rent-burdened households is quite significant: about half of renter households nationally pay more than 30% of their income on rent, and 27% – more than 12 million households – pay more than 50% of their income on rent.⁴ In New York City almost 30% of renters are severely rent-burdened, defined as paying more than 50% of their income on rent.⁵ 

However, measures of rent burden generally do not capture doubled up households. Estimates suggest that close to 4 million people live doubled up nationally, and the Coalition for the Homeless estimates that about 200,000 New Yorkers are doubled up.⁶ In New York City, estimates suggest that more than 150,000 school-aged children experienced homelessness during the 2024-2025 school year; this analysis uses a definition of homelessness that includes doubled up students.⁷ By contrast, there are approximately 20,000 school-aged children in the DHS shelter system, suggesting as many as approximately 130,000 New York City schoolchildren living doubled up.  

Because these doubled-up households are hard to find and quantify, they are very often ineligible for support programs, which are usually tied to indicators such as rent burden, shelter utilization, or formal eviction risk. Because those living doubled up are administratively invisible, they fall through the social safety net. Policymakers working to prevent and manage homelessness must think beyond rent burden and include doubled-up households in their projections of need and in the universe that should be eligible for safety net resources. 

2. Eviction and rent burden are not the primary drivers of homelessness. 

Attention to doubled-up households is particularly important because they are more likely to end up in shelter than someone with their own lease, even if the person with a lease is rent-burdened. This pattern holds nationally; the federal Department of Housing and Urban Development reports that only 9.8% of families entering shelter between October 2021 and September 2022 did so from their own permanent housing.⁸ At the same time, 29.1% of the families entering shelter were previously staying with family or friends. 

In New York City in FY 2023, only around 11% of families with children entering shelter reported doing so primarily as a result of eviction.⁹ High rent burdens and evictions are problematic in their own right but are not typically the factors that push households into homelessness.

Rental subsidies and other interventions are often targeted towards those at risk of eviction. In New York City, FHEPS is a State/City funded rental subsidy for families on Cash Assistance and who have been evicted or are at risk of eviction.  CityFHEPS is the City-funded rental assistance program and for those not in shelter is largely available only to households at risk of eviction.¹⁰ The City’s extensive Right to Counsel program for low-income households in Housing Court is frequently considered to be a homelessness prevention program. These programs are important and provide valuable supports for low-income households at risk of eviction. But if we want to shrink the shelter census, we must think beyond eviction prevention and focus on the population that is already homeless and that is living doubled-up with family or friends but is at risk of homelessness.

3. Low income households are resilient.

When the COVID pandemic started, I anticipated a significant spike in the number of families entering shelter.  Doubled-up living situations are challenging at the best of times, and more so when everyone is constantly in the home. Instead, family intake dropped substantially, resulting in a significant dip in the shelter census. In January 2020, there were 59,703 people on average daily in DHS shelter, including 37,433 people in families with children. In June of 2021, the shelter census was 47,669, of whom 26,663 were part of families with children (a 20.2% decline in overall shelter census and a 28.8% decline in individuals in families with children from January 2020).¹¹ 

During the pandemic, the federal government provided unprecedented levels of income supports: stimulus checks, enhanced SNAP benefits, expanded child tax credits, enhanced unemployment benefits, and others. These resources made it financially viable for households to stay together despite job losses and other pressures created by the pandemic.¹²

This evidence that households can and will stay together when it is financially viable to do so is critically important. Beyond the “natural experiment” of income supports provided during COVID, there is substantial evidence that social capital can prevent homelessness as long as the people in an individual’s network themselves have access to resources.¹³ Conversely, generations of systemic discrimination have limited wealth creation in communities of color, contributing to the over-representation of Black and Latino households among those experiencing homelessness.¹⁴ Creating public policies and programs that support household resiliency and wealth creation and foster personal safety nets will keep people from becoming homeless.

4. Prevention is essential.

As the COVID data above illustrate, reducing the number of people entering the shelter system has a direct impact on shelter census. Conversely, increasing exits out of shelter into permanent housing, while valuable for those moving, has a more muted effect on the overall shelter census. Over the last three City fiscal years, record numbers of households moved from shelter into permanent housing: in FY23 there were approximately 14,900 moves (a 17% increase from the prior year), in FY24 18,600 households exited to housing (a 25% increase), and in FY25 about 22,900 households moved to permanent housing (a 23% increase). Yet despite these strong exit trends, the shelter census essentially stabilized during that time period.¹⁵ Helping households move to permanent housing is necessary but not sufficient if the goal is to reduce the number of people experiencing homelessness.

Homelessness prevention is extraordinarily difficult to do well. The overwhelming majority of low-income and/or rent-burdened households will not experience homelessness. The sector has struggled to identify reliable indicators to predict shelter entry. Prevention needs to be both effective – successfully helping people avoid homelessness – and efficient – reaching primarily those who would have otherwise entered shelter. There are very few models that are successful on both dimensions.¹⁶ 

While difficult, homelessness prevention is an area that is ripe for innovation. We have more data and stronger technology than we have had in the past. Government, academia, and service providers should come together to build tools to better predict who will enter the shelter system. 

This collaboration could involve looking at specific sub-populations, such as pregnant people or older adults, as a starting point. While the drivers of homelessness writ large are immensely complicated, focusing on specific groups can help us isolate predictive factors. These sub-populations likely have different precipitating events, like a pregnancy or a medical emergency, that tip them from tenuous, often doubled-up living situations to homelessness. Deeper study of these sub-population using administrative data may help us better identify the factors that predict an individual will tip into homelessness.  

And the predictive model is only the first step in preventing homelessness. Once we can realistically anticipate who will enter the shelter system, we will need to design and implement targeted intervention programs. This effort will have to be an iterative process shaped by whether and how far upstream high-risk households can be identified. If we can identify these households, we will need the political will to force cross-sector collaboration, and make resources available.

5. We cannot ignore housing supply.

There is a clear relationship between housing markets and the incidence of homelessness: cities with constrained supply and high housing costs have higher rates of homelessness.¹⁷ But beyond this correlation, the availability of housing also has enormous impacts on the success of intervention programs.  

Rental vouchers are frequently cited as a key solution to homelessness, but vouchers only work if households can find a place to use them.¹⁸ Nationally, fewer than 60% of Section 8 voucher holders are able to successfully use their vouchers.¹⁹ While differences between the programs mean that analogous statistics do not exist for CityFHEPS in New York City, at any given point in time, there are 11,000-13,000 households in shelter with “shopping letters” searching for apartments on the private market.  

Vouchers are a go-to solution to homelessness for policymakers because programs can be stood up relatively quickly with low upfront administrative costs. And for the households that are able to successfully lease up, a voucher is often the key to stable housing. In New York City, families with children exiting the shelter system to subsidized housing have only a 1% return rate.²⁰  

But voucher programs alone are insufficient to seriously address homelessness. Large-scale investment in housing supply so that people exiting shelter can use  their vouchers is also critically important. Traditionally, homeless services are considered the purview of human services agencies and are often divorced from the affordable housing community. We have to build connections between sectors and maximize development opportunities that serve the lowest income households. Project-basing vouchers can be an effective tool to do so; there is a long history of project-basing Section 8, although limited funding jeopardizes public housing authorities’ ability to do this going forward.²¹ New York City has just begun to project-base CityFHEPS; there is significant opportunity to do more here to create new units that will serve the formerly homeless.

Vouchers are virtually always restricted to traditional household structures.  One family holds both the voucher and the lease. There are strict occupancy standards, and changes in family composition can make a household ineligible for their apartment. Shared housing arrangements are typically not allowed. Given the significant absolute shortage of housing in New York City and in many other communities across the country, making vouchers more flexible to permit alternative living arrangements could serve as a functional expansion of housing supply and better allow voucher holders to serve as the safety net for their extended networks.  

6. Income matters.

Housing affordability is a function of both how much housing costs and how much people have to spend, but efforts to improve housing affordability almost always focus exclusively on housing cost. This imbalance has resulted in a situation where even those living in subsidized housing can be at risk of becoming homeless. Nationally, 40% of households living in housing financed through the Low Income Housing Tax Credit Program pay more than 30% of their income in rent.²² In New York City, the New York Housing Conference, using NYU Furman Center data, found that subsidized housing accounted for more than a third of eviction filings in 2024.²³

Not surprisingly, households experiencing homelessness are overwhelmingly very low income.  In New York City, for example, approximately 85% of the families in the shelter system receive cash assistance.²⁴  While cash assistance rules and associated income limits vary, the New York City “standard of need,” considered to be the baseline income standard for families on cash assistance, is $789/month for a family of 3, or $9,468 annually.²⁵ 
These families may exit shelter to subsidized housing, but they remain at risk for future homelessness because their income is still so low.  

Investment in policies and programs that both increase wages and income supports for very low-income households and help these household members get and sustain jobs and then advance along career paths is an important part of addressing housing instability. 

7. The relationship between homelessness and public safety is very complicated.

There is an almost universal assumption that homelessness leads to community safety risks. Indeed, the President of the United States released an Executive Order about homelessness in July of 2025 that opens by stating that  “endemic vagrancy, disorderly behavior, sudden confrontations, and violent attacks have made our cities unsafe.”²⁶ While homelessness – particularly unsheltered homelessness – is challenging for communities for many reasons, this narrative that homelessness causes violent crime does not hold up.

There is ample evidence that people experiencing homelessness are frequently victims of crime; a study of people experiencing homelessness in San Francisco, for example, found that 32.3% of women, 27.1% of men, and 38.1% of transgender persons reported a history of either sexual or physical assault in the previous year.²⁷ Data show that people experiencing homelessness are far more likely to be victims of violent crime than people who are housed.²⁸ 

People experiencing homelessness commit crimes – in New York City, for example, rates of arrest among people experiencing unsheltered homelessness are far higher than among the general population.²⁹ 
The majority of these individuals were arrested just once, typically for larceny or low-level drug crimes. Of course there are also people experiencing homelessness who do commit violent crimes. But to reduce homelessness to a public safety issue that impacts only housed people in the community misses the public safety issue and real risks to unhoused people and leads us to miss opportunities to improve safety for everyone.  

Shelter is safer than the streets or subways. The number of violent incidents in the adult shelter system in New York City stands at 2.8 per 1,000 people, down 32% since 2023.³⁰Helping people come inside and connecting them with wrap-around services is good for both individuals and the communities that then face less unsheltered homelessness, and studies have found that the presence of a shelter does not change local crime rates.³¹ Rather than accepting the rhetoric that homelessness makes people unsafe and using that as justification for opposing homeless services, we need to invest in the programs – including shelter – that make us all better off.

8. Homelessness is not going away in the near term.

The United States absolutely can and should take steps to address homelessness, and, as I have discussed above, there are many opportunities for reducing the number of people experiencing homelessness. But the numbers of people either already experiencing homelessness or at risk of doing so are enormous.

While ambitious goals like ending homelessness are necessary to drive large-scale change, that messaging has also resulted in decades of short-term and expensive decision making. If we truly believe that homelessness will end soon, there is no reason to invest in long-term solutions, like high-quality shelter, housing supply, job training and prevention, that are necessary to support people who may be at risk of homelessness down the road. This orientation is particularly evident in the real estate choices made around programs that serve people experiencing homelessness. Shelters are often in buildings rapidly – and poorly – adapted from other uses, putting clients and shelter staff in challenging conditions. In many cases, jurisdictions use hotels as shelters. Even though New York City has the most robust shelter infrastructure in the country, it still uses more than 16,000 hotel rooms as shelter. Not only do hotels lack the appropriate social service and common space, but they are also an extremely expensive option.

Had the City acknowledged a generation ago that homelessness is a large-scale and long-term problem, it could have avoided hundreds of millions of dollars of spending while providing higher quality services. Recognizing the scope of the challenge is not capitulation. Ending homelessness is a goal worth fighting for, but we need infrastructure robust enough to meet people’s needs in the meantime.  

I firmly believe that we can meaningfully reduce the number of people experiencing homelessness. But the traditional toolbox of federal homeless assistance programs is shrinking, and need is growing. We cannot count on Continuum of Care-funded supportive housing for single adults or Section 8 vouchers for families, and even at the best of times, those resources have been in scarce supply. Learning from my experience – and that of other practitioners – will be critical as we shape the next generation of policies and programs.