HUD Data Show Housing Voucher Costs Leveled Off Starting In 2003 As Rental Market Cooled
August 23, 2005
The Administration has pointed to "spiraling" costs for the Housing Choice Voucher program to justify radical changes in the program's funding structure and the elimination of many key protections for low-income families. HUD and Treasury data indicate, however, that far from spiraling, growth both in the average cost of a voucher and overall expenditures under the Section 8 program (which includes the voucher program and a separate "project-based" assistance program) have eased considerably since peaking in 2003. The housing voucher program helps about two million low-income households -- most of them senior citizens, people with disabilities, and working families -- to rent modest housing in the private market.
The recent slowdown in spending growth reflects the fact that most of the increase in program costs in 2003 and preceding years was driven by temporary pressures that have since eased.
- A "perfect storm" of a hot housing market and a cooling economy resulted in a temporary uptick in voucher costs. A voucher covers the difference between the cost of a modest apartment and about 30 percent of a family's income, so if the gap between rents and incomes widens rapidly, voucher costs will rise as well. During 2001 and 2002, that gap grew more rapidly than during any period in more than 30 years. Rental cost increases slowed beginning in 2003, however, and average incomes among the poorest fifth of households leveled off after dropping in 2001 and 2002.
- Voucher costs also rose because Congress chose to expand the voucher program; few new vouchers have been added since 2002. About 53 percent of the growth in Section 8 expenditures from 1996 to 2003 resulted from new vouchers that Congress created to assist families at risk of losing other forms of federal housing assistance or to reduce the number of families on waiting lists.
Part of the slowdown also may reflect a shift that HUD made in April 2004 in the method for distributing voucher funds to state and local housing agencies. The slowdown in average voucher costs started in the summer of 2003, however, and the majority of the decline occurred before April 2004, so the funding shift was not the main cause of the slowdown. Instead, the slowdown appears to show that cost-control mechanisms built into the voucher program worked as planned.
Voucher Cost Growth Has Fallen Below Rate of Inflation
Recently released HUD data show that the average cost of a housing voucher rose by about 2.1 percent from the November 2003-January 2004 quarter to the November 2004-January 2005 quarter. By comparison, the overall rate of inflation in the Consumer Price Index during this same period exceeded 3 percent. According to the HUD data, which housing agencies submit to HUD each quarter, growth in per-voucher costs peaked at a quarterly rate of 2.6 percent in February-April 2003 and has since declined for seven consecutive quarters beginning with May-July 2003. 
Treasury data show that overall spending under the Section 8 program also slowed sharply in 2004. Changes in overall Section 8 spending may differ from changes in average voucher costs, because Section 8 includes a separate "project-based" housing assistance program in addition to the voucher program and because the figures reflect changes in the number of households receiving assistance in addition to changes in average costs per voucher. The rate of increase in overall Section 8 spending fell from 13.2 percent in fiscal year 2003 to 6.7 percent in fiscal year 2004.
Based on examination of trends in rents, incomes, and the number of vouchers in use, the Congressional Budget Office (CBO) projected in March 2004 that growth in Section 8 spending would slow to 1.8 percent in fiscal year 2005.  CBO's analysis preceded changes HUD made to the voucher funding system in April 2004 and was based on the assumption that Congress and HUD would provide housing agencies with adequate funds to support all authorized vouchers in use.
Claims that Section 8 is Consuming the HUD Budget Rely on Misleading Budget Authority Data
Critics of Section 8 have claimed that rapid growth of the program as a proportion of the HUD budget threatens to result in reduced funding for other HUD programs. The claim that the Section 8 program is growing rapidly as a share of the HUD budget, however, rests on misleading use of "budget authority" figures.
The federal government measures budget trends in programs in two ways: by tracking changes in the amounts of new funding (or "budget authority") that Congress appropriates for a program, and by tracking changes in the level of actual program expenditures (or "outlays"). The two measures paint sharply different pictures of growth in the Section 8 program in recent years. From 1996 to 2004, budget authority for the program grew at an annual rate of 18 percent, while outlays grew at an annual rate of 4 percent. As a result, the share of HUD discretionary budget authority allotted to Section 8 rose from 26 percent in 1996 to 55 percent in 2004. But the share of HUD outlays spent on Section 8 has remained nearly steady (fluctuating between 50 and 57 percent).
The important figures here are those for outlays. This is true for two reasons. First, it is outlays -- or actual government expenditures -- that determine a program's cost and its impact on budget deficits or surpluses. Second, the unusually large increases in budget authority for the Section 8 program are an artifact of a major change that was made in the Section 8 funding structure. Most Section 8 units were initially funded in the 1970s and 1980s through long-term contracts under which Congress provided up-front all of the budget authority expected to be needed to support the units during the full term of the contracts, which usually was a period of several decades. Beginning in the mid-1990s, these long-term contracts started to expire, and Congress began providing new funding for these units on an annual, rather than a multi-year, basis. As a result, the amount of new budget authority needed to support the existing Section 8 units each year grew at a substantial clip. But this increase in budget authority did not reflect an increase in program costs or in the level of assistance provided.
Outlays for Section 8 units are recorded each year as they occur, regardless of whether the unit is under a long-term contract or a one-year contract. Outlay growth provides the only accurate measure of actual increases in Section 8 costs or in the share of HUD spending that is accounted for by Section 8.
Cost Growth Slowed as Temporary Pressures Relented
Growth in Section 8 spending has slowed in large part because most spending growth through 2003 was driven by temporary pressures that have since eased. Approximately 53 percent of the growth in Section 8 expenditures from 1996 to 2003 stemmed from Congressional decisions to increase the number of families that the program assists. Approximately 33 percent of the growth reflected growth in average costs of existing vouchers required to keep pace with a widening gap between housing costs and the incomes of voucher holders.  It is likely that some increases in spending in recent years also resulted from successful efforts to reduce the number of housing vouchers left unused. Each of these three pressures abated to some degree in 2003 and 2004.
Number of New Vouchers Has Declined
The creation of new housing vouchers has been the largest source of growth in Section 8 spending since 1996. The number of vouchers authorized by Congress rose by a total of approximately 657,000 from 1996 to 2003.
In recent years, Congress has expanded voucher program enrollment in two main ways. First, Congress has provided vouchers to families that are losing certain other types of federal housing subsidies -- such as public housing, assistance under federal mortgage subsidy programs, and project-based Section 8 assistance -- as a consequence of Congressional decisions to allow those subsidies to end.  These vouchers account for more than two-fifths of the total number of new vouchers provided from 1996 to 2003. Because they took the place of existing subsidies, these vouchers did not result in a net increase in the number of families receiving housing assistance, and the resulting growth in voucher costs is offset to some degree by a reduction in expenditures elsewhere in the budget.
Second, Congress created some new "incremental vouchers" between 1999 and 2002. These vouchers helped to reduce the number of eligible families unable to participate in the voucher program -- and often stuck on long waiting lists -- because of funding limitations. Only about one in every four low-income families eligible for vouchers receives any type of federal housing assistance.
No new incremental vouchers have been authorized since 2002. As a result, the overall increase in the number of vouchers has dropped from a peak of about 130,000 vouchers in 2001 to around 30,000 a year from 2003 to 2005, and every one of the vouchers added from 2003 to 2005 will go to replace a housing subsidy that was previously provided under another federal program but will be eliminated.
This decline did not immediately cause spending growth to slow, because it takes some time after Congress appropriates funds for additional vouchers for HUD to award the vouchers to state and local housing agencies and for the agencies to distribute the vouchers to families. New vouchers typically are not in use until the year after the year for which Congress approves them, and then, quite frequently, only for part of that year. For example, the new vouchers approved in the fiscal year 2002 appropriations act did not result in increases in expenditures until fiscal year 2003, and it was not until fiscal year 2004 that the increased costs showed up for all 12 months of the year. Because of this two-year phase-in, the diminished growth in the number of authorized vouchers starting in 2003 only began to result in slower growth in voucher spending in 2004 and its impact will not be fully felt until 2005. Indeed, 2005 likely will be the first year since 1998 in which no costs are incurred for new "incremental" vouchers.
Growth in Gap Between Rents and Incomes Has Slowed
The first years of this decade saw private-market rent and utility costs rise at an unusually rapid pace, while the earnings of low-income families declined as the result of the economic downturn. In 2001 and 2002, rent and utility costs rose by 8.0 percent, while average incomes among the bottom fifth of households fell by 1.6 percent. The incomes of voucher holders would be expected to fluctuate somewhat less in response to economic changes than the incomes of low-income families generally, because about 40 percent of voucher holders are elderly or have disabilities and receive much of their income from public pensions that grow based on cost-of-living indices, rather than employment or wage levels. HUD data on the incomes of voucher tenants for 2002 and 2003 -- which because of data and administrative lags reflect actual incomes of tenants beginning in 2001 -- show growth of 1.4 percent a year (well below the rate of housing cost growth) compared to increases of more than three percent in each of the preceding five years.
Since vouchers typically cover the difference between the cost of a modest apartment and 30 percent of a family's income, the government's cost in providing a voucher rises when the gap widens between the rental charge and a family's income. Increases in the average voucher subsidy needed to keep pace with the growing gap between market rents and the incomes of voucher holders accounted for about 33 percent of the total growth in Section 8 costs from 1996 to 2003.
The combination of a hot rental market and a cool economy in 2001 and 2002 represented an unusual "perfect storm." There has been no period in more than 30 years when trends in rental housing costs and incomes of low-income families diverged so markedly. This trend has since subsided. During 2003, residential rents grew at their slowest pace since 1997, while incomes of low-income families leveled off.  Rents still rose somewhat more in 2003 than the incomes of low-income families, so the gap between housing costs and incomes continued to edge up, but not nearly as rapidly as in 2001 and 2002.
As with the slowdown in the number of new vouchers created, the full effect of this change on voucher costs was not felt immediately. This was the case because voucher program rules set caps (referred to as "payment standards") on the amount of rent a voucher can cover. These caps are not instantly adjusted in response to changes in market rents, so they can have the effect of delaying the impact of market trends on voucher costs.
The extent of this delay will vary from one family's voucher to another. Rents covered by vouchers used by families living in housing units with rents below the cap can rise freely as long as they remain below the cap level (and the housing agency does not find the new rent to be out of line with the local market) so the effect of market rent trends on the cost of these particular vouchers would be felt more or less immediately. For families living in units with rents at (or above) the cap, however, the impact is felt more slowly. Payment standards are set based on local "Fair Market Rents" that HUD adjusts each year, using housing market data that are usually months old. In addition, after HUD adjusts a Fair Market Rent, other administrative processes may result in it taking another year or more for the adjustment to be reflected widely in the rent-subsidy levels of individual families.
As a result, the more rapid rent increases that occurred in 2001 and 2002 continued to influence voucher cost growth at least into fiscal year 2004. Similarly, effects of the cooling of the rental market nationally in 2003 and 2004 will not be fully reflected in voucher program costs until at least 2006.
Growth in Proportion of Authorized Vouchers in Use Slowed
Each year, some families issued vouchers are unable to use them, often because they cannot find an apartment they can rent with the voucher. The share of authorized vouchers that state and local housing agencies actually put to use rose from 91 percent of such vouchers in fiscal year 2001 to approximately 96 percent in fiscal year 2003.  This increase reflects the success of efforts by HUD and state and local housing agencies to raise the proportion of authorized vouchers that actually are in use serving families. These efforts were undertaken after Congress expressed strong concerns that too many authorized vouchers were being left unused, and were facilitated by a number of voucher program reforms that Congress encouraged or required and that HUD instituted.
State and local housing agencies cannot receive funding for more than 100 percent of their authorized vouchers and, as a practical matter, there is virtually no chance that most housing agencies will be able to put 100 percent of their vouchers in use. All of this means that it simply is not possible for the rate of increase in voucher utilization seen in 2002 and 2003 to continue.
Timing of Slowdown Suggests Existing Market-Based Cost Controls -- Rather than a New "Fixed-Cost" Funding System -- Played Key Role
On April 22, 2004, HUD announced a change (described in greater detail below) in the system for distributing voucher funding to state and local housing agencies that was designed to restrain growth in voucher costs. By April, however, nearly half of the slowdown in growth of average voucher costs from 2.6 percent in the quarter from February-April 2003 to a slightly negative rate of growth in the quarter from November 2004 to January 2005 had already occurred. This indicates that the new funding system was not the primary cause of the slowdown.
Instead, it appears that cost controls already built into the voucher program ensured that spending growth slowed as the housing market cooled and temporary cost pressures let up. A series of program rules tightly link spending to the number of vouchers authorized by Congress, market housing costs, and the incomes of tenants:
- Each state or local housing agency has a limited allotment of vouchers, and it cannot obtain funds to issue vouchers beyond that allotment. 
- The amount of rent that a voucher can cover is capped by the local "payment standard," which is based on "Fair Market Rents" that HUD establishes for more than 2,000 local markets, based on the typical rental charge for a modest apartment in the area.
- State and local housing agencies are required not only to enforce the overall rent cap, but also to compare the rent of every apartment to be occupied by a voucher-holder to rents for similar unsubsidized apartments in similar locations, and to refuse to allow vouchers to be used for apartments where the rent is out of line with the local market. The only HUD study examining implementation of this "rent reasonableness" rule found that the requirement was generally followed and that average voucher rents were on average $95 per month below rents for similar unsubsidized units.
- Voucher tenants are required to contribute about 30 percent of their income toward rent, reducing the cost of their voucher to the government.
Cost Control Structure Remained in Place After 1998 Policy Changes
Critics of the voucher program have argued that controls on voucher costs were effectively eliminated by policy changes made in the late 1990s. They have pointed in particular to a 1998 legislative change allowing housing agencies to set voucher payment standards up to 10 percent above or below the HUD Fair Market Rent. Previously, housing agencies were not permitted to set payment standards above the Fair Market Rent anywhere in the geographic area that the Fair Market Rent covered, even though many such geographic areas included a number of different housing submarkets and some had populations as large as several million people. The added flexibility provided by the 1998 law was intended to allow agencies to set payment standards at a level that met the needs of their communities and to permit voucher holders to choose to live outside of the neighborhoods with the lowest rents -- which also tend to have higher poverty rates, higher crime, and fewer education and employment opportunities.
It is possible that by allowing payment standards above the Fair Market Rent, the added flexibility contributed somewhat to growth in voucher costs. HUD has reported that the average payment standard in 2003 was 104 percent of the Fair Market Rent. This indicates that housing agencies overall had not pushed payment standards close to the limit of 110 percent, but nonetheless that agencies, on average, had made more use of flexibility to raise maximum rents above the FMR than they made of flexibility to set them below the FMR. It may be, however, that much of this payment standard increase was needed simply to keep pace with market housing costs. Since increases in Fair Market Rents lag behind growth in market housing costs -- especially during hot housing markets like the one that occurred during the first years of the decade -- many agencies likely used their added payment standard flexibility to make more timely adjustments to respond to increases in market rents, rather than to increase voucher rents over and above the growth in market costs.
Moreover, an increase of one percent in the payment standard does not necessarily cause voucher costs to be one percent higher, since many voucher holders rent apartments with rents below the payment standard, and consequently their subsidy levels are not affected by payment standard changes. A 2002 HUD-sponsored study was unable to find any significant effect on voucher costs stemming from payment standards in the range from 101 to 110 percent of the local FMR, but that report was based on data from a relatively small sample of housing agencies. 
As noted above, our analysis of voucher costs since 1996 found that growth in the number of vouchers and the widening gap between overall market rents and the incomes of voucher holders accounted for 86 percent of the growth in Section 8 expenditures from 1996 to 2003. An additional 9 percent stemmed from three other factors: growth in administrative funding (which appears to have been well below the overall rate of inflation), a small amount of growth in project-based Section 8 per-unit costs, and increases in funding for several small sub-programs that are included in the Section 8 budget account. It is likely that most of the final 5 percent of Section 8 cost growth reflects growth in voucher rents beyond the rate of growth in overall market rents as reflected in the Consumer Price Index. This could in part reflect the impact of payment standard flexibility, but there are several other reasons why it would be expected that voucher rents would rise somewhat faster than the CPI:
- Additional vouchers were directed to areas with higher costs. The vouchers that were newly put to use (either through expansion in the number of authorized vouchers or growth in the share of existing vouchers that were actually used) from August 2002 to July 2004 went to agencies that, on average, had per-voucher costs 8 percent above the national average.
- Newly leased apartments tend to have higher rents than apartments where a family has lived for some time. Landlords tend to raise rents more when they rent an apartment to new tenants than when tenants remain in the same apartment. As a result, rents for newly-leased apartments tend to be higher than for other apartments. As noted above, from 1996 to 2003, the voucher program expanded substantially. In addition, from 2001 to 2003, as many as 100,000 previously unused vouchers were put to use. As a result, particularly during the later part of this period, the proportion of voucher holders who had recently entered the program was unusually high. Some of these families used their voucher to rent the unit where they were already living, but many moved to new units.
- Many vouchers issued to replace other federal subsidies were subject to special rules that prevented tenant displacement but allowed higher rents. When owners of federally subsidized apartment buildings decide to stop accepting federal project-based subsidies and to increase the rents for units in the buildings, low-income tenants who reside in these buildings can face displacement from their homes. To prevent that from occurring, Congress established rules under which tenants in these buildings can receive a special type of voucher. Under the rules that govern these "enhanced" vouchers, the rent levels that the vouchers can cover may be set at the levels charged for comparable units, even if such rent levels exceed the rental charges that vouchers normally can cover. Most of the tenants who receive enhanced vouchers are low-income elderly people or people with disabilities.
- Rents for modest apartments may grow at different rates than rents for higher-end units. CPI rental cost data cover the full housing market, including many high-end rental units that would not be available to families with vouchers. If rents for mid- and low-end housing units that voucher holders are more likely to rent grow at a different pace than high-end units, the CPI data will not accurately reflect rents of units available to voucher holders.
Regardless of what role payment standard flexibility and other factors played in voucher costs, it is important to note the added flexibility was a one-time adjustment that Congress made to the voucher program's cost controls in order to reduce the concentration of voucher holders in the highest poverty neighborhoods and to promote other program goals it considered important. Any cost increase stemming from that change, which began to be implemented in 1999, should now be completed and is unlikely to affect current or future cost trends. Since housing agencies are still required to set maximum voucher rents within federally established limits and to ensure that every rent paid by a voucher is in line with the local market, the overall structure of market-based cost controls remains firmly in place.
"Fixed Cost" Budgeting Provides Added Layer of Cost Containment but Places Low-Income Families at Risk
Although much of the slowdown in voucher costs discussed in this paper preceded the April 2004 changes in HUD's method for providing funding to state and local housing agencies, it is likely that those funding changes have further ratcheted down voucher costs to some degree. A portion of that impact would not show up in the voucher data used in our analysis, because the data cover the period only through January 2005 and not all of the impact of policy changes that housing agencies made in response to the new funding method would have been felt by that date. 
Before 2004, HUD generally provided each agency with enough funding to support all of the agency's vouchers that were in use at their actual cost. Under the fiscal year 2004 funding policy announced in April, HUD instead provided agencies with enough funds for all vouchers in use at their actual cost in May-July 2003 plus a formula adjustment based on projections of rental inflation in the region in which the housing agency is located. In cases where the amount of funding provided under HUD's formula was below the actual cost of vouchers, housing agencies (unless they happened to have access to funding reserves sufficient to cover the shortfall) were required to reduce spending under their voucher programs.
Many housing agencies found themselves in this situation, because the new formula did not match funding closely to actual needs. HUD's rental inflation projections covered areas including as many as eight states, so HUD often applied an identical inflation adjustment to scores of local markets facing widely different trends in rent and utility costs. Moreover, the formula entirely ignored the impact on voucher costs of flat or declining tenant incomes and a number of other factors (such as changes in the average size of voucher households in a local community or moves by voucher tenants from high-poverty communities to communities with lower poverty but somewhat higher rents). 
In August 2004, HUD provided supplemental funds beyond the amount set by the new formula in response to appeals by some underfunded housing agencies. But many agencies facing shortfalls did not submit appeals -- in part because HUD did not establish criteria for the appeals process in a clear and timely manner -- and often agencies that did appeal for added funding did not receive the full amount they needed. Moreover, by the time HUD got around to awarding the additional funds, many agencies had already reduced spending through measures that could not easily be undone or that agencies were reluctant to undo out of concern that they would be shortfunded again in 2005.
This policy constrained voucher spending growth by forcing local housing agencies to cut back expenditures, but it did so at considerable cost to low-income families in some areas. To reduce their expenditures to the amount covered by the new funding formula, many housing agencies had no choice but to cut back assistance to low-income families. Many agencies reduced the number of vouchers in use by "shelving" vouchers that became available when families left the program on their own instead of issuing the vouchers to new families on the waiting list. Other agencies took steps to reduce the average cost of their vouchers, for example by imposing new minimum monthly payments or lowering maximum voucher rents. In a few cases, agencies took vouchers away from families that were already using the vouchers to help pay the rent.
Voucher Cost Growth Will Remain Moderate
Growth in voucher costs will remain slow in 2005, and it is probable that this slow growth will continue in fiscal year 2006 and beyond. Continued slow growth would be likely even if Congress and HUD had provided full funding in 2004 and 2005 to cover the actual costs of vouchers in use. This is the case because the underlying conditions that prompted rapid growth in voucher spending into 2004 -- approval of a large number of new vouchers and a "perfect storm" of surging rents and slumping incomes -- are not expected to return.
The voucher funding provisions adopted by Congress in the fiscal year 2005 appropriations act, however, will cause the slowdown in cost growth to be deeper and more abrupt than it would have been as a result of these "natural" trends in voucher costs. The act instructed HUD to continue the fixed cost system adopted in 2004, with three significant changes:
- First, according to current HUD estimates, Congress only provided 96 percent of the funds that agencies would be entitled to in 2005 under the fixed cost formula contained in the 2005 appropriations act. This shortfall occurred partly because the estimates of program funding needs that HUD provided to Congressional appropriators turned out to be inaccurate. As a result, many agencies will be forced to make cuts in voucher assistance even in cases where the amount calculated under the formula would have been adequate to cover the agencies' costs.
- Second, Congress prohibited HUD from providing housing agencies with additional funds to cover increases in the number of vouchers in use or rent increases exceeding HUD's regional inflation factor. In 2004, by contrast, agencies were able to obtain funds to cover growth in the utilization of authorized vouchers and, as noted above, HUD made adjustments to its inflation factor to reflect local variations in rental cost increases, in response to appeals from agencies. Though these adjustments were late and often inadequate, they nonetheless reduced the extent of the cuts that agencies were forced to make.
- Third, Congress sharply reduced the amount of funding agencies have access to through contingency funds (known as "program reserves"). In 2004, these funds served as a crucial resource for agencies that received inadequate funding under HUD's formula. 
The Administration's fiscal year 2006 budget would restore some of the vouchers cut in 2005, but the budget also calls for much larger cuts in later years. In addition, an Administration proposal introduced in Congress in April would convert the voucher program to a block grant and weaken or eliminate key protections for low-income families -- steps that would make the proposed cuts easier to implement.  Critics of the voucher program have argued that these cuts and changes are necessary to bring voucher cost growth "under control." This argument ignores the reality that the period of rapid voucher cost growth has already ended and that the cost containment measures in place in 2003 and 2004 proved themselves up to the task of managing program spending.
Funding Method and Cost Containment Incentives Should Be Improved
This is not to say that the existing cost containment provisions should not be improved. The rigid, fixed-cost funding structure used in 2004 and 2005 is an unnecessarily blunt instrument that ratchets down spending even when the consequences for low-income families and landlords are harsh and when the causes of cost increases are clearly beyond the control of state and local housing agencies. Even if Congress maintains a structure that bases funding on voucher costs in the past adjusted for inflation (rather than on actual costs), a few key improvements -- such as altering the inflation adjustment to take income data into account and be more sensitive to local rent trends and providing housing agencies with adequate contingency funds that are replenished annually -- could significantly reduce such a structure's adverse effects without retreating from the Congressional goal of providing an added layer of cost containment beyond that offered by market-based controls.
In addition, the market-based cost-control mechanisms could themselves be fine-tuned. HUD's process for setting Fair Market Rents results in voucher payment standards in some areas that are higher than is needed to enable voucher holders to rent housing, while in other areas, it causes payment standards to be too low. And additional HUD monitoring of compliance with the requirements that rents paid by vouchers be reasonable in the context of local markets might further improve enforcement of that mandate.  In addition, certain changes to the rules for calculating tenant contributions to rent could simplify administration and further reduce errors without raising rent burdens on needy families.
Housing Choice Vouchers have been demonstrated to be a highly effective form of housing assistance. The bipartisan, Congressionally-mandated Millennial Housing Commission described the voucher program as "flexible, cost-effective, and successful in its mission" in its 2002 report, while a 2002 study by the General Accounting Office found the program to be more cost-effective than other major federal housing programs. A 2003 report by the Office of Management and Budget on the performance of HUD programs states that the voucher program demonstrates "improved efficiencies and cost effectiveness in achieving program goals each year" and that independent evaluations "indicate that the program is effective and achieving results." And as recently as November 2004, HUD's Fiscal Year 2004 Performance and Accountability Report found that the voucher program succeeded in meeting eight of the nine performance goals that HUD had established since 2001. Those goals covered areas including expanding use of voucher to support homeownership and encourage self-sufficiency, and reducing errors in subsidy calculations.
As this analysis of Section 8 spending growth in recent years indicates, most of the growth has been driven by a combination of rising housing costs and Congressional decisions to expand the program, rather than by an erosion of program efficiency. Careful steps to improve further the voucher program's cost-effectiveness would be worthwhile. There is, however, no need for drastic cost-cutting measures that risk harming the program and the approximately two million low-income households -- most of them working families, elderly people, and people with disabilities -- that it serves.
 The per-voucher cost figures used in this analysis are calculated based on HUD Voucher Management System (VMS) data, which are used to determine voucher funding levels for most housing agencies. They exclude 16 agencies that received funding in 2004 or before through special arrangements under the Moving-to-Work demonstration and were not required to report VMS data for most of the quarters covered by our per-unit cost analysis.
 CBO has not published an updated version of this analysis since 2005. CBO's January 2004 budget estimates projected Section 8 outlays using a different method that did not take into account rents, incomes, or the number of vouchers in use.
 Two types of tenant-based assistance -- Section 8 tenant-based certificates and Section 8 vouchers -- were merged into the current Housing Choice Voucher Program by the Quality Housing and Work Responsibility Act (QHWRA) of 1998. As used in this analysis, the term "voucher" includes both pre-QHWRA vouchers and tenant-based certificates and post-QHWRA vouchers.
 The methodology used in this analysis is described in " Sources and Methods Used to Estimate Components of Section 8 Cost Growth ."
 New vouchers were awarded to replace some, but not all, of the 191,000 project-based Section 8 units that were lost from 1996 to 2003. The 53 percent share of Section 8 cost growth that resulted from the increase in the number of units is the net increase in costs resulting when the cost of new vouchers is offset by savings from the loss of project-based Section 8 units.
 These figures are based on mean bottom quintile income data from the Current Population Survey and a weighted composite of the Consumer Price Indices for residential rents and for fuel and utilities. The residential rent CPI alone rose by a total of 8.6 percent in 2001 and 2002. The statement elsewhere in this analysis that the divergence of rents and incomes in 2001 and 2002 was the largest in more than 30 years is based on analysis of these same data sources for each calendar year since 1967.
 In inflation-adjusted terms, incomes among bottom-quintile households continued to drop in 2003. According to Current Population Survey data, however, the average unadjusted bottom-quintile income (which is the proper basis for comparison with rent inflation and voucher costs) remained nearly unchanged, rising by 0.06 percent compared to 2002.
 In the years prior to 2003, HUD provided some funds to state and local agencies for vouchers that did not end up being put to use that year. These funds were counted as outlays, even though they were not actually spent on vouchers and were later recaptured by HUD. Because this accounting practice artificially inflated outlays reported for 2002 and before above the real cost of vouchers in use, only part of the increase of approximately 5 percent in voucher utilization from 2001 to 2003 was reflected in increased outlays.
 Before 2003, agencies that used more than their allotted number of vouchers were able to use voucher contingency funds to cover the resulting costs, but then had to reduce the number of vouchers in use below their allotment number to pay back the costs stemming from this "overleasing." In 2003, Congress banned agencies from using more than their allotted number of vouchers over the course of a fiscal year, even if they reduced utilization the following year. Temporary overleasing was a useful management tool for state and local housing agencies, because the number of families that are able to use their vouchers successfully can be unpredictable. Agencies were able to overissue vouchers in the same manner that airlines overissue tickets, which allowed a larger share of agencies' vouchers actually to be used, and helped compensate for the fact that a substantial number of families issued vouchers end up being unable to use them.
 Quality Control for Rental Assistance Subsidies Determinations, Final Report, Office of Policy Development and Research, HUD, June 20, 2001, p 64.
 Finkel et al, Costs and Utilization in the Housing Choice Voucher Program, Abt Associates, July 2003 . The study did find that setting payment standards above 110 percent of the FMR, which the 1998 law allows housing agencies to do under certain circumstances with HUD approval, was linked to significant cost increases at agencies taking this measure. However, such "exception payment standards" are rare. Data provided by HUD in January 2004 showed payment standards above 110 percent of the FMR were in effect in areas containing 2.5 percent of the nation's population.
 The fiscal year 2004 appropriations act funded vouchers during calendar year 2004, so the period directly affected by the act had concluded by January 2005. Many of the policy changes housing agencies made from April 2004 to December 2004 in response to shortfalls, however, will continue to suppress cost growth after January 2005. For example, some agencies reduced voucher payment standards as a way of covering their funding shortfalls, but due to administrative lags the full impact of a payment standard change on voucher costs would not be felt for a year or more after the change was made.
 For further information on cuts carried out by housing agencies in response to HUD's 2004 funding policy, see Local Consequences of HUD's Fiscal Year 2004 Voucher Funding Policy, CBPP, July 15, 2004.
 For additional information on the fiscal year 2005 funding policy, see Barbara Sard, Peter Lawrence, and Will Fischer, Appropriations Shortfall Cuts Funding for 80,000 Housing Vouchers This Year, February 11, 2005.
 See Will Fischer and Barbara Sard, Administration Housing Proposal Lays Groundwork for Planned Funding Reductions, May 9, 2005.
 HUD has indicated that it plans to intensify efforts to improve enforcement of rent reasonableness requirements in 2005.