Getting better outcomes with less public spending is always desirable, and our current fiscal situation adds urgency to this task. Low-income housing assistance is fertile ground for such reforms. Most current recipients are served by programs whose cost is enormously excessive for the housing provided. Phasing out these programs in favor of the system’s most cost-effective program would ultimately free up resources to provide housing assistance to millions of additional people while reducing taxpayer cost. It would also reduce other deficiencies of the current system.

The current system provides large—sometimes enormous—subsidies to some households while offering none to others that are equally poor, and it provides subsidies to many people who are not poor while offering none to many of the poorest. Finally, the current system is strongly biased against homeownership for low-income households. A modest modification of the system’s most cost-effective program would eliminate this bias.

Overview of Current System

To see how to get better outcomes for less money spent on low-income housing assistance, it is essential to understand the nature of current programs and the evidence about their performance.

The bulk of low-income housing assistance in the United States is funded by the federal government through a number of programs at a combined cost of over $50 billion a year. No single program dominates the system. Unlike other major means-tested transfer programs in the United States, low-income housing programs do not offer assistance to many of the poorest families who are eligible for them. Eligible families that want assistance must get on a waiting list for a housing program when it is open for new applicants.

Most U.S. low-income housing assistance is for renting a unit. The most important distinction between rental housing programs is whether the subsidy is attached to a housing project (project-based assistance) or the assisted household (tenant-based assistance).

Tenant-based assistance / Each family offered tenant-based assistance is free to occupy any unit that meets the program’s minimum housing standards, rents for less than the program’s ceiling applicable to the family, is affordable with the help of the subsidy, and whose owner is willing to participate in the program. The family retains its subsidy if it moves to another unit meeting those conditions.

The U.S. Department of Housing and Urban Development’s Housing Voucher program is the only significant program of tenant-based assistance. It is the largest low-income program, serving about 2 million households and accounting for about 30 percent of all households that receive low-income rental assistance.

Project-based assistance / Under programs of project-based assistance, each family must accept the particular unit offered in order to receive assistance, and loses the subsidy if the family moves from the housing project unless the family is able to obtain alternative housing assistance before moving.

There are two broad types of project-based rental assistance: public housing and privately owned, subsidized projects. Both types have usually involved the construction of new projects; in almost all other cases, they have required substantial rehabilitation of existing buildings. Many of these programs no longer subsidize the construction of projects, but most projects built under them still house low-income households with the help of subsidies for their operation and renovation. Overall, project-based assistance accounts for about 70 percent of all households that receive low-income rental assistance.

Public housing projects are developed and operated by local public housing authorities established by local governments, albeit with substantial federal subsidies and regulations that restrict the local governments’ choices. In the public housing program, government employees make most of the decisions made by unsubsidized for-profit firms in the private market: what to build, how to maintain it, and when to tear it down. Decisions about where to build projects have been heavily influenced by local political bodies. The public housing stock has declined by about 300,000 units since its peak in 1991. About 1 million households live in public housing projects.

Government agencies also contract with private parties to provide housing in subsidized projects. Most are for-profit firms, but not-for-profits have a significant presence. This is the largest part of the system, involving a number of different programs. The largest programs of this type are the Internal Revenue Service’s Low-Income Housing Tax Credit (LIHTC), HUD’s Section 8 and Section 236 project-based programs, and the U.S. Department of Agriculture’s Section 515 and Section 521 programs. Under these programs, in exchange for certain subsidies, private parties agree to provide rental housing meeting certain standards at restricted rents to eligible households for a specified number of years. None of the programs provide subsidies to all suppliers who would like to participate, and this is highly relevant for their performance. About 4 million households live in projects of this type.

Program Performance

Of all the differences between the various housing programs, differences in cost-effectiveness are, by far, the largest and most consequential for housing policy. Differences in most other aspects of performance across the programs are relatively small.

Evidence on the performance of housing programs indicates that project-based assistance has a much greater cost than tenant-based assistance when it provides equally good housing. These studies define equally good housing to be housing that would rent for the same amount in the unsubsidized market in the same locality. This measure accounts for the desirability of the neighborhood as well as the housing itself. In the best studies, the estimated magnitude of the excess cost is enormous.

The best study of HUD’s largest programs that subsidized the construction of privately owned projects (the Section 8 New Construction and Substantial Rehabilitation programs) found that the total cost of providing housing under those programs was at least 44 percent greater than the total cost of providing equally good housing under the voucher program. This implies that it would have been possible, using housing vouchers, to serve all of the people served by the New Construction and Substantial Rehabilitation programs equally well and serve at least 72 percent more people with the same characteristics without any increase in public spending. The best study indicates even larger excess costs for public housing. In contrast, several studies over the years have found that the total cost of various types of tenant-based housing assistance has exceeded the market rent of the units involved by no more than the cost of administering the program. Landlords receive market rents for their units. The voucher program is also the least expensive to administer.

More recent evidence has confirmed the large excess cost of the New Construction and Substantial Rehabilitation programs. Based on a large random sample of HUD-insured Section 8 projects, Meryl Finkel et al. found that in 1995, property owners received payments from tenants and the government that exceeded market rent for 86 percent of their units. Two-thirds of the units received payments that exceeded market rent by more than 20 percent. The true excess cost of those projects was even greater because the predicted market rents were based on the assumption that all property systems are restored to their original working condition and hence overstate the market rents of the units in their current condition. The calculations also ignore other subsidies received by these projects when they were built. Mark Shroder and Arthur Reiger, in a 2000 Journal of Housing Research paper, produced similar results in a direct comparison of the total payments to property owners under the Section 8 Rental Certificate program, a tenant-based program, and the Section 8 New Construction and Substantial Rehabilitation programs. They found that New Construction and Substantial Rehabilitation units are about 35 percent more expensive than comparable units rented under the Certificate program. U.S. General Accounting Office studies in 2001 and 2002 produced similar results for the major active construction programs LIHTC, HOPE VI, Section 202, Section 515, and Section 811.

Incentives and good decisions / What accounts for the large differences in the total cost of providing equally good housing under programs of tenant-based and project-based assistance? Although evidence about the relative importance of various factors does not exist, the plausible explanations are the absence of a financial incentive for good decisions on the part of civil servants who operate public housing, the excessive profits that inevitably result from allocating subsidies to selected developers of private, subsidized projects, the resources that developers devote to securing the subsidies, and the distortions in usage of inputs resulting from the subsidy formulas. A special case of the latter is that project-based assistance is usually tied to the construction of new units. The least-expensive approach to improving the housing conditions of low-income households involves heavy reliance on upgrading the existing housing stock, the primary mechanism through which tenant-based assistance achieves this goal.

The results concerning the cost-effectiveness of different housing programs illustrate the virtue of substantial reliance on market mechanisms for achieving social goals, especially the virtue of forcing sellers to compete for the business of buyers. Under a program of tenant-based assistance, only suppliers who provide housing at the lowest cost given its features can remain in the program. If the property owner attempts to charge a voucher recipient a rent in excess of the market rent, the tenant will not remain in the unit indefinitely because he or she can move to a better unit without paying more for it. The subsidy is portable. Under programs of project-based assistance, suppliers who receive payments in excess of market rents for their housing can remain in the program indefinitely because their tenants would lose their subsidies if they moved. These suppliers have a captive audience.

The difference in the performance of the USDA’s Section 502 and HUD’s Section 235 homeownership programs lends support to the importance of this factor. The primary difference between these programs concerned the allocation of the subsidies. Families offered assistance by the USDA program were free to buy any house that met the program’s standards and the families didn’t face any significant time constraint for doing it. In the largest component of the HUD program, the subsidies were allocated to selected builders, and eligible families had to buy the housing units from those builders. Unlike the USDA program, the HUD program suffered from scandals and high default rates that resulted from above-market prices paid for the houses.

Tenant-based assistance has another major advantage over project-based assistance. It greatly expands recipient choice. All low-income housing programs have minimum housing standards that reflect the interests of taxpayers in ensuring that low-income families live in adequate housing. Project-based assistance forces each family to live in a particular unit in order to receive a subsidy. As a result, it greatly restricts recipient choice among units meeting minimum housing standards. In contrast, each family offered tenant-based assistance has a choice among many units that meet the program’s standards, and the family can retain its subsidy when the family moves. The landlord does not have a captive audience.

Excessive subsidies / Unlike other major means-tested transfer programs, housing assistance does not offer assistance to all families that are eligible for it, despite its stated goal of “a decent home and suitable living environment for every American family.” Millions of the poorest families are not offered any housing assistance, while a smaller number of equally poor families receive large subsidies. For example, an assisted family with one child and an adjusted annual income of $12,000 living in an area with the program’s average payment standard would have received an annual housing subsidy of about $7,800 from the Housing Choice Voucher program in 2014 if the family occupied an apartment renting for at least the payment standard. The majority of families with the same characteristics living in that locality would receive no subsidy from any low-income housing program. Furthermore, the majority of the poorest eligible families receive no assistance while many families with considerably greater income are assisted. This is not because the poorest families do not want assistance on the terms offered; the waiting lists of public housing authorities are long, would be much longer in most cases if they were open continuously for new applicants, and consist largely of families with extremely low incomes.

An Oct. 9, 2013 segment on PBS News Hour illustrates the excessive subsidies. It revealed that $500,000 per apartment had been spent building a housing project for the homeless in San Francisco. This is expensive even by that city’s standards. The median value of owner-occupied houses in the San Francisco metro area was $558,000 and the median household income of their occupants was $104,000. So this government program provided apartments to the poorest families that were almost as expensive as the houses occupied by the average homeowner. Insuring that the homeless occupy housing meeting reasonable minimum standards does not require anything like the amount of money spent on these units. More than 20 percent of owner-occupied houses in the area sold for less than $300,000. Furthermore, almost half of the families in the area were renters whose median income was about $50,000. They lived in much less expensive units than homeowners on average. In Portland, Ore., where the median value of owner-occupied houses was $249,000, $360,000 per apartment was spent to build another housing project for the homeless.

These cases are not anomalies; they are repeated throughout the country. The desire of the people involved in the current system to provide the best possible housing for their clients is understandable. However, this is a luxury that we can ill afford. We do not need to build new units to house the homeless; they can be housed in satisfactory existing units at a much lower taxpayer cost. More than 6 percent of the dwelling units in the San Francisco area were vacant at the time that the project for the homeless was built.

Plausible assumptions about taxpayer preferences argue strongly for replacing the current patchwork of low-income housing programs with a housing assistance program that offers the same assistance to all eligible families in the same circumstances. Surely, few favor offering some households large subsidies while denying assistance to other identical households. Instead, the public favors offering the same assistance to all households that are equally needy.

Discouraging homeownership / Finally, the current system of low-income housing assistance is heavily biased against homeownership. After accounting for geographical price differences and adding an imputed return on home equity to the income of homeowners in calculating their real income, nearly 25 percent of renters—but less than 5 percent of homeowners in the lowest real income decile—receive housing assistance. The gap in the second real income decile is smaller, but still substantial. Few argue that the government should actively discourage homeownership by low-income households. Low-income housing assistance should be at worst neutral in this regard. The flawed innovations in mortgage finance that sparked the recession of the late 2000s say more about how, rather than whether, to deliver housing assistance to low-income homeowners.

Getting More for Less

The available evidence on program performance has clear implications for housing policy reform. To serve the interests of taxpayers who want to help low-income families with their housing and the poorest families who have not been offered housing assistance, Congress should shift the budget for low-income housing assistance from project-based to tenant-based housing assistance as soon as current contractual commitments permit and should not authorize any new programs involving project-based assistance. This would increase the cost-effectiveness of the system. Moreover, replacing the current system of low-income housing assistance with a tenure-neutral housing voucher program that offers the same assistance to all eligible families in the same circumstances would eliminate its inequities and bias against homeownership.

With due attention to the transition to the new system, the benefits of the proposed reforms can be achieved without hurting current recipients. Indeed, a transition can be designed to benefit many of these households. For example, public housing tenants could be offered a choice between housing vouchers and staying in their current units on the same terms. Current recipients of Section 8 vouchers could be allowed to receive the generous subsidies that are now offered by the program while new recipients receive less generous subsidies so that more households can be served. Reform must also honor legal commitments. For example, payments on current terms will be provided to owners of private subsidized projects until the end of their use agreements. After the transition is complete, millions of additional families would receive housing assistance that enables them to occupy better housing in nicer neighborhoods and consume more of other goods. Millions of other families that would have received project-based assistance with the continuation of the current system would live in housing and neighborhoods that they prefer to their units in subsidized projects.

To illustrate the magnitude of the gains that would ultimately flow from these reforms and allay concerns about their effects on various subgroups, I have estimated the effects on the number of families served of phasing out the bulk of HUD’s programs in favor of alternative housing voucher programs that offer assistance to all of the poorest households. Most have slightly smaller taxpayer cost than the current programs, but one spends about 10 percent less. Those estimates are based on data from HUD’s administrative records and the 5 percent Public Use Microdata Sample from the 2000 Decennial Census. The participation rate in the proposed program for families of each type is based on experience from the only entitlement housing assistance program that has been operated in the United States: the Housing Assistance Supply Experiment. Like the Supply Experiment, the proposed programs provide a subsidy conditional on occupying a unit meeting certain minimum housing standards similar to the standards in the current housing voucher program. In most voucher programs considered, the magnitude of the subsidy is equal to a payment standard minus 30 percent of the recipient’s adjusted income. Payment standards are larger for families whose size and composition justify more bedrooms, and they are different in different localities. Most are designed so that recipients who occupy units renting for the local payment standard would live in equally good housing in all locations. In most, the upper-income limit for eligibility is the lowest income at which the subsidy is zero. Unlike the current system, income limits are the same in real terms in all localities.

The results of the analysis of the voucher program that costs 10 percent less are striking. It would serve 75 percent more people in total and many more in families of each type. The percentage increase is different for people of different types in part because some groups such as large families and Hispanics are served at much lower rates by the current system than others in the same economic circumstances. However, more people of each type would be served as a result of the reform—189 percent more Hispanics, 85 percent more whites, and 8 percent more blacks; 17 percent more elderly and 87 percent more nonelderly; 76 percent more living in metro areas and 71 percent more living in non-metro areas; 93 percent more in the first real income decile and 48 percent more in the second real income decile; and 17 percent more in one- and two-person households, 74 percent more in three- and four-person households, and 159 percent more in households of five or more persons.

Spending less but not helping less / When interpreting these results, it is important to distinguish between the amount of money spent on behalf of a recipient and how much the recipient is helped. Because less would be spent on a larger number of people, the amount spent per person would be less under the reformed housing voucher program than with the continuation of the current system. Because they would be grandfathered, current recipients would not be hurt by the reforms. However, many future recipients would have less money spent on their behalf than would have been spent with the continuation of the current system.

This does not imply that the recipients would be served less well. As mentioned earlier, the best previous studies show that housing vouchers could be used to provide occupants of subsidized housing projects with equally good housing for the same rent at a much lower taxpayer cost. The amount spent on behalf of recipients under different types of low-income housing assistance is a poor measure of how much they are helped by it. The 75 percent increase in the number of people served as a product of the reform is to a considerable extent, the result of replacing programs of project-based assistance with much more cost-effective housing vouchers.

This is not to say that all future recipients of low-income housing assistance would be better served by this reformed housing voucher program than the continuation of the current system. For example, the reformed program’s subsidies are less generous than in HUD’s Section 8 Housing Choice Voucher Program. This reduction in subsidies to future voucher recipients serves the interest of equity. Instead of providing larger subsidies to two identical households and none to a third, the proposed reform provides smaller subsidies to all three. Some believe that less generous voucher subsidies will preclude participation by the poorest households, but the results in this paper indicate that the poorest families will experience larger than average increases in participation because of the reforms.

Other studies using data from other sources and years have produced similar results. In a cruder version of the preceding analysis, based on the 2007 Current Population Survey, Dirk Early and I found that replacing the current system with an equally costly housing voucher program that offers assistance to all of the poorest households would increase the number of households served by at least 47 percent and reduce the poverty rate by 67 percent more than the current system. Because the proposed reform increases participation by the largest families at a higher than average rate, the percentage increase in the number of people served would be even greater. In preliminary calculations based on the 2011 American Housing Survey, Emily Snow and I found that replacing the current system with an equally costly housing voucher program that offers assistance to all of the poorest households would increase the number of people served by at least 113 percent and have a similar pattern of percentage increases across people of different types.

Objections to Reform

Some object to providing housing assistance to all eligible families or exclusive reliance on tenant-based housing assistance. This section addresses their concerns.

The main argument against offering housing assistance to all families that are eligible for it is that it would be too expensive. Delivering housing assistance to all currently eligible families using the current mix of housing programs would greatly increase the amount spent on housing assistance. However, we do not have to make more than a third of the population eligible for low-income housing assistance. Furthermore, we can reduce the fraction of housing assistance delivered through programs that are cost-ineffective and we can provide new recipients of housing assistance with smaller subsidies. If we reduce the fraction of the population eligible for housing assistance, increase the fraction of families served by tenant-based assistance, or reduce the subsidy to new recipients under each housing program, we can create a program that serves all of the poorest households that apply for assistance, and the program would cost less than the current system.

It is often argued that a shortage of affordable housing calls for subsidizing the construction of new units. This argument is seriously flawed. Almost all people are currently housed. If we think that their housing is too expensive (commonly called unaffordable), the cheapest solution is for the government to pay a part of their rent. The housing voucher program does that. This program also insures that its participants live in units that meet minimum standards. Building new units is a much more expensive solution to the affordability problem.

Furthermore, it isn’t necessary or desirable to construct new units to house the homeless. The number of people who are homeless is far less than the number of vacant units—indeed, far less than the number of vacant units renting for less than the median. In the entire country, there are only about 600,000 homeless people on a single night and more than 3.6 million vacant units available for rent. Even if all homeless people were single, they could be easily accommodated in vacant existing units, and that would be much less expensive than building new units for them. Besides, most of the 600,000 people who are homeless each night already have roofs over their heads. They live in a special type of subsidized housing called homeless shelters. The best provide good housing.

Finally, some critics of reform are concerned that tenant-based assistance will not work well in markets with the lowest vacancy rates because those markets do not have enough affordable vacant apartments that meet minimum housing standards to house all families who are offered vouchers. In fact, it is not necessary for the number of vacant apartments that meet minimum housing standards and are affordable to voucher recipients to exceed the number of new and recycled vouchers available in order to use all vouchers available. Many families that are offered vouchers already occupy apartments meeting the program’s standards. We do not need vacant apartments for those families; they can participate without moving. Other families that are offered vouchers live in housing that does not meet Section 8 standards; however, many of their apartments can be improved to meet the standards. Similarly, vacant apartments that do not initially meet the program’s standards can be upgraded to meet the standards. In short, a tenant-based voucher program increases the supply of apartments meeting minimum housing standards. Evidence from the tenant-based Section 8 Certificate and Voucher programs shows that these are the market responses to additional housing vouchers.

Conclusion

The major defects of the current system of low-income housing assistance are the cost-ineffectiveness of the programs that account for the majority of households served, the inequities associated with its failure to offer the same housing assistance to all eligible families in the same circumstances, and its bias against homeownership. Phasing out those programs in favor of the system’s most cost-effective program would ultimately free up the resources to provide housing assistance to millions of additional people and reduce taxpayer cost.

Building new units is an extremely expensive way to provide better housing to low-income households. Subsidizing selected suppliers to do it is especially expensive. Renting existing units that meet minimum standards is much cheaper. This also avoids providing recipients of low-income housing assistance with better housing than occupied by the poorest families that are ineligible for assistance. The proposed reforms will gradually move the system of low-income assistance toward more cost-effective approaches and enable us to provide housing assistance to millions of additional people without spending more money.

Readings

  • Federal Housing Assistance: Comparing the Characteristics and Costs of Housing Programs, GAO-02–76, produced by the U.S. General Accounting Office. 2002.
  • Federal Housing Programs: What They Cost and What They Provide, GAO-01–901R, produced by the U.S. General Accounting Office. 2001.
  • Final Comprehensive Report of the Freestanding Housing Voucher Demonstration, Vols. 1 & 2, by Mireille L. Leger and Stephen D. Kennedy. Abt Associates, 1990.
  • “Geographic Price Variation, Housing Assistance, and Poverty,” by Dirk W. Early and Edgar O. Olsen. In Oxford Handbook of the Economics of Poverty, edited by Philip N. Jefferson; Oxford University Press, 2012.
  • “Getting More from Low-Income Housing Assistance,” Discussion Paper 2008–13, by Edgar O. Olsen. Brookings Institution, Hamilton Project, 2008.
  • Housing Allowances and Other Rental Assistance Programs—A Comparison Based on the Housing Allowance Demand Experiment, Pt. 2: Costs and Efficiency, by Stephen K. Mayo, Shirley Mansfield, David Warner, and Richard Zwetchkenbaum. Abt Associates, 1980.
  • “Housing Programs for Low-Income Households,” by Edgar O. Olsen. In Means-Tested Transfer Programs in the U.S., edited by Robert Moffitt; University of Chicago Press, 2003.
  • Participation and Benefits in the Urban Section 8 Program: New Construction and Existing Housing, Vols. 1 & 2, by James E. Wallace, Susan Philipson Bloom, William L. Holshouser, and Shirley Mansfield. Abt Associates, 1981.
  • “Promoting Homeownership among Low-Income Households,” Report No. 2, by Edgar O. Olsen. Urban Institute, Opportunity and Ownership Project, 2007.
  • Quality Control for Rental Assistance Subsidies Determination, produced by ORC/​Macro. U.S. Department of Housing and Urban Development, Office of Policy Development and Research, 2001.
  • Status of HUD-Insured (or Held) Multifamily Rental Housing in 1995: Final Report, by Meryl Finkel, Donna DeMarco, Deborah Morse, Sandra Nolden, and Karen Rich. Abt Associates, 1999.
  • “The Cost-Effectiveness of Alternative Methods of Delivering Housing Subsidies,” by Edgar O. Olsen. Unpublished manuscript, University of Virginia, 2009.
  • “The Effect of Fundamental Housing Policy Reforms on Program Participation,” by Edgar O. Olsen. Unpublished manuscript, University of Virginia, 2014.
  • “The Performance and Legacy of Housing Policies,” by Edgar O. Olsen and Jens Ludwig. In The Legacies of the War on Poverty, edited by Martha Bailey and Sheldon Danziger; Russell Sage Foundation, 2013.
  • “United States Housing Policy,” by Edgar O. Olsen and Jeffrey E. Zabel. In Handbook of Regional and Urban Economics, Vol. 5, edited by Giles Duranton, J. Vernon Henderson, and William Strange; North-Holland, forthcoming.
  • “Vouchers versus Production Revisited,” by Mark Shroder and Arthur Reiger. Journal of Housing Research, Vol. 11, No. 1 (2000).