literature review

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=rhpd20 Download by: [University of Utah] Date: 21 July 2017, At: 13:56 Housing Policy Debate ISSN: 1051-1482 (Print) 2152-050X (Online) Journal homepage: http://www.tandfonline.com/loi/rhpd20 Are low-income housing tax credit developments locating where there is a shortage of affordable units?

Kirk McClure To cite this article: Kirk McClure (2010) Are low-income housing tax credit developments locating where there is a shortage of affordable units?, Housing Policy Debate, 20:2, 153-171, DOI:

10.1080/10511481003738260 To link to this article: http://dx.doi.org/10.1080/10511481003738260 Published online: 13 May 2010.Submit your article to this journal Article views: 345View related articles Citing articles: 7 View citing articles Are low-income housing tax credit developments locating where there is a shortage of aﬀordable units?

Kirk McClure* Department of Urban Planning, University of Kansas, Lawrence, KS, USA Low-Income Housing Tax Credit (LIHTC) developments serve renter households with incomes between 30% and 60% of Area Median Family Income. Ideally, the program places units into neighborhoods where there is a shortage of units serving this cohort. LIHTC units are allocated to developers by state agencies through their Qualiﬁed Allocation Plans which should direct units to areas of need. Using a national database, this research examines where LIHTC developments were placed in service to determine whether these developments enter tracts experiencing shortages.

The LIHTC program is not directing units to those census tracts where there is a latent demand for units in this rent range. Rather, it is placing units into tracts that have surpluses. Equally, the program is not placing units in tracts with little or no aﬀordable housing. This suggests that the program is not breaking down the income separation that exists in the nation’s housing markets.

Keywords:low-income housing; tax credit; rental housing; aﬀordability Introduction The Low-Income Housing Tax Credit (LIHTC) program is now the nation’s largest program designed to produce aﬀordable rental housing for low-income households.

The nation’s stock of LIHTC units, through 2007, is about 1.5 million units in over 28,000 projects (US Department of Housing and Urban Development 2008b). Each year, developers in each state prepare proposals for projects which compete to receive tax credits awarded by each state housing ﬁnance agency. The state agencies rank order the proposals according to criteria published in the agency’s Qualiﬁed Allocation Plan (QAP) with the top ranked proposals winning the tax credits.

Through this process, an average of almost 1400 projects and 100,000 units were placed in service each year during the period from 1995 to 2004. At this pace of production, the LIHTC program either has surpassed or will soon surpass all the other forms of project-based housing assistance for low-income renter households. It has already grown larger than the slowly contracting supply of public housing with its 1.0 million units. It is also larger than any of the individual project-based subsidy programs that are no longer growing such as the Section 236 and Section 8 New Construction and Substantial Rehabilitation programs. Finally, the LIHTC *Email: [email protected] Housing Policy Debate Vol. 20, No. 2, March 2010, 153–171 ISSN 1051-1482 print/ISSN 2152-050X online 2010 Virginia Polytechnic Institute and State University DOI: 10.1080/10511481003738260 http://www.informaworld.com program is also on track to surpass the Housing Choice Voucher program with its 1.9 million units (Haley 2008, 1).

This pace of production also means that the LIHTC program units comprise a signiﬁcant share of all rental units being built. Nationwide, about 420,000 rental units, both subsidized and unsubsidized, were built each year during the 1990s.

During the same time period, the LIHTC program placed in service about 72,000 units per year. About two-thirds of the LIHTC units are newly constructed, while the remaining units were developed through the rehabilitation of existing units.

Thus, the newly constructed LIHTC program subsidized about one in every eight of all rental units being built. Despite this share, the program may generate no net increase in the total stock of rental housing due to a high degree of substitution. For nearly every unit added through the LIHTC program, one unsubsidized unit may have been displaced (Malpezzi and Vandell 2002). The LIHTC program serves households of low-income, while the total stock of rental housing serves renters households of all income levels. The program may simply direct more new growth into a segment of the market serving households with lower income than might have been the case otherwise. The program may also bring aﬀordable rental housing into neighborhoods, especially suburban neighborhoods, where, due to higher develop- ment costs or local opposition, it might not have been possible to develop rental housing aﬀordable to low-income households in the absence of the tax credits (McClure 2006).

This large scale program consumes a great deal of public resources, estimated at $5.3 billion in 2008 (Burman 2008), and these resources should be used as eﬀectively as possible. To achieve this desired level of eﬀectiveness, the program should be placing units into markets where there is a shortage of rental units aﬀordable to low- income households. This paper examines this issue through two research questions.

First, is the LIHTC program placing units in markets where there is a shortage of units in the price range served by the program? Second, is the LIHTC program placing units in markets where few or no aﬀordable housing options existed before, even if an area has no apparent latent demand for aﬀordable units?

The implications are straightforward. If the program is adding units where there is latent demand for the increased supply of housing, then the program is working eﬀectively. If not, then the program is adding units where an adequate supply already exists. Latent demand is deﬁned as a market condition where the number of low-income households in the tract is greater than the count of rental units aﬀordable to this group. Where latent demand exists, there is an identiﬁable need for the units subsidized by the LIHTC program. However, latent demand may not exist.

Adding subsidized units to this type of market with an already saturated supply can create downward pressure on prices and push vacancy rates up. This hurts property values and dampens investment in the maintenance and improvements of the properties (Knaap 1998). Adding units to an already saturated market can also attract additional low-income households to this market, increasing the concentra- tion of poverty which works against the nation’s eﬀorts to disperse the poor (Downs 1981).

Literature review The LIHTC program was created as part of the Tax Reform Act of 1986. This program was not carefully designed to oﬀer a speciﬁc form of capital subsidy for the 154K. McClure production of aﬀordable rental housing. Rather, the program was hastily created out of a desire to restore some form of subsidy to aﬀordable housing in the wake of the Congress’ eﬀorts to eliminate a variety of tax provisions that had favored all rental housing and were deemed to be ineﬃcient (Buron et al. 2000; McClure 2000). The use of tax credits, as opposed to a direct capital grant for housing production, has been criticized as ineﬃcient and overly generous to developers (Case 1991; Stegman 1991). Others have applauded the program for its ﬂexibility and increasing cost eﬀectiveness (Cummings and DiPasquale 1999). Despite this origin and its contested design, there is agreement on one aspect of the program; it is meant to help resolve the shortage of aﬀordable rental housing units for low-income households (Abravanel and Johnson 2000).

Although the program is directed toward the development of aﬀordable units serving low-income renter households, it is important to understand that it does not serve the poorest of the poor. Cummings and DiPasquale (1999, 278–79) found that, ‘‘The income limits under the LIHTC program of 50 and 60% of area median income results in rents that are beyond the means of many poor households.’’ Generally, the program serves some of the least worst oﬀ among those of low income. Nelson (1994, 411) states, ‘‘LIHTC occupants must have incomes between 40 and 60% of the median to avoid severe rent burdens, and research shows that families who occupy such units do have incomes in that range.’’ The program sets a ceiling on the income of households who reside in the assisted units, but in practice, property owners ﬁnd it necessary to attract those tenants whose income are close to, but not above, the income ceilings set for the program.

Each development receiving the tax credits must adhere to the ‘‘20/50–40/60’’ rule.

This rule states that at least 20% of the units must be set aside for households with income below 50% of the applicable Area Median Family Income (AMFI) or at least 40% of the units be set aside for households below 60% of AMFI. 1Setting the minimum for the percentage of units set aside for low-income occupancy has not had much meaning as a vast majority of all LIHTC developments have 100% of the units occupied by households meeting the selected income limitations (McClure 2006).

Because the LIHTC provides exactly the same beneﬁts to a development whether it opts for the higher 60% of AMFI limit or the lower 50% of AMFI limit, a overwhelming majority of developments select the higher limits and most developments have rents at or very close to the maximum allowed (Buron et al.

2000).

Rent limits accompany these income limitations. These rent limits are derived from HUD’s published area median family income levels and enforced within each state by the agency that allocates the tax credits among developments. The rents are not tied to the income of the resident; rather, they are based upon the income of a household in the metropolitan area who meets the program’s income limitations.

Rent maximums are set annually for each metropolitan area and non-metropolitan county from HUD’s published income limits. The rents are calculated beginning with the applicable AMFI which is multiplied by 60% (or 50% depending upon the income ceiling selected) to determine the income ceiling for the target population. 1The US Department of Housing and Urban Development publishes the Area Median Family Income and the resulting income limits for the LIHTC program for each metropolitan area and non-metropolitan county in the nation. These can be found at: http://www.huduser.org/ datasets/il.html. Housing Policy Debate155 This is then multiplied by a housing burden percentage ranging from 35% for eﬃciency apartments to as much as 64% of units with ﬁve bedrooms. 2This ﬁgure is divided by 12 to get the monthly gross rent for each size unit in each part of the nation. The unit’s contract rent is derived by taking this gross rent and reducing it by a utility allowance, usually taken from the local public housing authority, for any utilities paid by the tenant (Guggenheim 2003, 22–23).

At the discretion of the ownership, rents actually charged in the development may be lower than the maximum rents permitted. Frequently, LIHTC developments are found in neighborhoods that confront concentrated poverty and very low market rents. This can create market pressure to keep rents in the LIHTC developments below the maximum permitted levels as the rents in competing properties may be very low. Thus, the LIHTC development may be at the top of the range of rents charged in the neighborhood, which may be less than the maximum rents permitted by the program (Buron et al. 2000).

Given these rent and income limitations, the LIHTC program tends to serve households whose income places them in the 30% to 60% of AMFI. Because rents are set assuming that the resident household is at the maximum income permitted (50% or 60% of AMFI) with a housing cost burden that is high (35% or more) property managers seek households at or near the top of the permitted income levels.

If a poorer household is placed in the unit, the likelihood of a rent collection problem increases. The program provides no incentive to help poorer households, and the workings of the program create an incentive for property managers to house those households with the highest permitted incomes (McClure 2006).

Little is known on the actual incomes of the households served by the program.

Owners of properties in the LIHTC program are not required to report tenant information to either the federal government or to the state agency administering the program. Owners are required to submit to periodic inspections of their properties by the state agency, but these inspections are designed to enforce the ceilings on allowable household income and on unit rents. The inspections are not designed to collect data, such as tenant income level, for research purposes.

However, a few studies have reported on the income levels for samples of LIHTC households. Ernst & Young (1997) reported on a large sample of properties and found that, for those households not receiving voucher assistance, the average household occupying a LIHTC unit had income placing it at 45% of the AMFI. This study also found that about one-third of the households received Section 8 voucher assistance permitting even poorer households to aﬀord the rents in the LIHTC developments. Thus, layering the voucher subsidy on top of the LIHTC subsidy permits the LIHTC program to serve households who live below poverty. In the absence of this additional voucher subsidy, it appears that the LIHTC program serves households who are low-income but are above the poverty level. An Abt Associates study in 1999 examined 39 properties in ﬁve cities (Buron et al. 2000).

This study found 40% of the households had income below 30% of AMFI, which is approximately equivalent to the poverty level. However, the Abt study also found that the majority (54%) of the households had income between 30% and 80% of AMFI. It found that the typical LIHTC households had income above the median household income in the area. Finally, the Government Accounting Oﬃce made a 2At present, there are no regulations as to the maximum or minimum size of a household for a particular units by bedroom size (Guggenheim 2003, 21).

156K. McClure comparable ﬁnding in 1996 that incomes of LIHTC households are at or above the median for households in the area (US Government Accounting Oﬃce 1997). Note that this is household income, not the higher family income used to set the income and rent limits. Thus, the typical LIHTC household has an income placing it above the medianhouseholdincome in the area, which generally means about 45% of the area medianfamilyincome and falling between 30% and 60% of the AMFI.

By comparison, analysis of the national data on Housing Choice Vouchers (formerly known as Section 8 vouchers) ﬁnds that the voucher program serves households with income at 22% of AMFI (McClure 2005b). The voucher program is designed and targeted to serve the poorest of the poor. 3There is no minimum income in the voucher program, and there is an eﬀective ceiling on household income at about 30% of AMFI. In the LIHTC program there is a ceiling of 60% of AMFI and an eﬀective ﬂoor at approximately 30% of AMFI due to the pressure on property managers to lease units only to households who can aﬀord the rents. Thus, the LIHTC program is helping low-income households, but these households possess incomes that are much higher than those households helped by the Housing Choice Voucher program, the nation’s largest tenant-based rental assistance program.

The LIHTC program has been criticized for producing units for households who fall into this 30% to 60% of AMFI category (Nelson 1994). This is a market sector that often has adequate units, while households with extremely low income, those below 30% of AMFI, cannot aﬀord the rents in LIHTC developments. Nelson found that extremely low-income households must rely upon the private market to provide them with housing in markets that nearly always suﬀer from inadequate supplies of housing aﬀordable to this group. Nelson argued that the LIHTC program is not well designed to add units to the market segments that are short in supply. Rather, she found that shortages exist only at the rent levels that are less than one-half of the rents found in the LIHTC program and that the program tends to expand the supply of rental housing where expansion is not widely needed.

Finally, the LIHTC program has been critically examined in terms of where it is locating units. Freeman (2004) ﬁnds that LIHTC developments tend to be located in neighborhoods that are more racially and ethnically diverse, have lower concentra- tions of poverty, and enjoy greater increases in home values than those neighborhoods where other project-based assisted housing is located. He compared neighborhoods where LIHTC developments were built to those of other project- based housing programs. He found that the LIHTC units were disproportionately located in neighborhoods where these indicators of neighborhood quality were lower than for the population as a whole but higher than those found for the other project- based subsidy programs. He concluded that the neighborhoods where LIHTC developments located were not as economically disadvantaged as those with traditional federally assisted housing developments. Thus, whatever the level of need for the housing, developments in the LIHTC program are entering better neighborhoods than developments assisted through earlier vintage programs.

Comparing the LIHTC program to the Housing Choice Voucher program, the project-based program is credited with making entry into the racially and economically integrated suburbs at a rate equal to or greater than the Housing 3Each housing authority administering the Housing Choice Voucher program must provide 75% of its vouchers to applicants whose incomes do not exceed 30% of the area median family income (US Department of Housing and Urban Development 2008a). Housing Policy Debate157 Choice Voucher program which is designed to promote this type of spatial mobility (McClure, 2006). While this represents some level of improvement by the LIHTC program over other programs, much more needs to be done. Even where low-income households are able to obtain access to better neighborhoods, the LIHTC and other programs too often contribute to the social isolation of the poor and minorities.

(Freeman 2006).

While the LIHTC program may be able to locate in neighborhoods with higher measures of neighborhood quality than other project-based programs, problems remain; the LIHTC units seem to locate in clusters (Oakley 2008). The presence of one LIHTC project increases the likelihood of other nearby projects, and this could lead to new concentrations of poverty in the receiving neighborhoods. This clustering may be a result of a great many factors. Only limited sites may be oﬀered for sale in a market. Only limited sites may be available that do not confront prohibitive political obstacles to the development of low-income housing. Only limited sites may have the appropriate zoning for this type of housing. In addition, there is one feature of the program that gives greater subsidy to projects located in Qualiﬁed Census Tracts (QCTs). These QCTs are tracts with a high incidence of low- income households (at least 50% of households have an income less than 60% of the AMFI or the poverty rate is 25% or more). Such rewards cause the LIHTC developers to concentrate within these impoverished areas.

With this background, the national database of LIHTC units is examined to determine whether or not the program is successfully directing units into areas where there is a shortage of housing for the target population.

Is there an overall shortage of housing units for the target households?

Not much has changed since Nelson performed her 1994 study. The American Housing Survey for 2005 ﬁnds that there is a shortage of units only for households with the lowest income (US Census Bureau, American Housing Survey 2005). See Table 1. The survey reports that there were about 12.5 million renter households with income below 30% of AMFI. This group is known as the extremely low-income population. A household at the 30th percentile in 2005 had an income of about Table 1. Renter households by income level and rental units aﬀordable to these households.

Year 2000 2005 Renter households (000s) Income below 30% of AMFI 9,913 12,504 Income between 30% and 60% of AMFI 9,659 9,869 Rental units with rents aﬀordable to households (000s) Income below 30% of AMFI 6,769 5,839 Income between 30% and 60% of AMFI 20,659 17,676 Surplus or deﬁcit (000s) Income below 30% of AMFI73,14476,665 Income between 30% and 60% of AMFI 11,000 7,807 Source: US Bureau of the Census, Census 2000.

US Bureau of the Census, American Housing Survey 2005.

158K. McClure $17,400. These households could aﬀord gross rents (contract rent plus tenant-paid utilities) of no more than $435 per month. There are only 5.8 million rental units in the nation renting below that level, far less than the number of extremely low-income households. This results in a shortfall in 2005 of about 6.7 million rental units.

Households between 30 and 60% of AMFI are the ones served by the LIHTC program. These households had income from $17,400 to $34,800 in 2005. 4The survey found about 9.9 million such renter households. These households can aﬀord gross rents in the range of $435 to $870 per month. There were about 17.7 million units in this price range, many more than the number of renter households who could aﬀord these units. Thus, a large surplus exists, about 7.8 million units.

Note that the surplus in the rental housing stock aﬀordable to those in the 30th to 60th percentile households is greater than the shortfall for those below the 30th percentile. This suggests that there is no overall shortage of lower priced housing for all households below 60% of AMFI. Rather, the distributions of low-income renter households by income and rental housing units by rent levels are shaped diﬀerently.

The rental units are disproportionately found in the upper reaches of the rent distribution while the renter households are disproportionately found toward the lower end of the income distribution.

A surplus does not mean that the units are available for rent. Nothing stops households with a higher income from renting these units, removing them from the market for these low- and very low-income households. This approach to measuring a surplus is only an approximation to the matchup between supply of and the demand for rental housing aﬀordable to those households with income between 30% and 60% of AMFI. A surplus nationally also does not mean that this surplus condition exists in all markets. Some market areas may have surpluses, reﬂecting the national conditions. Other market areas have shortages of units serving the 30% to 60% of AMFI households.

The units developed with assistance from the LIHTC program are expected to be placed in markets where there is a demonstrated need for housing. The state agencies awarding tax credits to developments must approve a comprehensive market study of the housing needs of low-income renters in the area. These studies are usually prepared by the developer or agent working for the developer. The state agency must make a ﬁnding that there is a need for this additional housing in the local market to be served by the proposed units (Guggenheim 2003, 69). This procedure is designed to direct the LIHTC units to just those areas where a demonstrated need for such housing exists.

Is the LIHTC program sending units to areas with a shortage?

Ideally, the LIHTC program is placing units into neighborhoods where there is a latent demand for such housing. This is the purpose of the market analysis requirement imposed on all developers’ applications for tax credits. Latent demand is demand for rental housing that is not served by the existing supply of rental housing. The latent demand for aﬀordable rental housing is estimated for all census 4These ﬁgures correspond to 30% and 60% of the national median family income. The estimated number of households is approximate because adjustments could not be made for household size. Housing Policy Debate159 tracts across the country for the year 2000, using decennial census data. 5It is calculated as the number of renter households whose income falls between 30% and 60% of the AMFI for the metropolitan area or non-metropolitan county minus the number of rental units aﬀordable to these households. The analysis then determines the extent to which the LIHTC program located units in the subsequent years to resolve the problems arising from this demand in excess of supply. The analysis also examines the subset comprising renter households with income below 30% of AMFI, the group generally too poor to aﬀord LIHTC housing.

At the national scale, the rental housing market conditions in the year 2000 were similar to the conditions found in 2005 (see Table 1). In 2000, there were 9.7 million renter households with income below 60% but above 30% of the applicable AMFI.

That year, there were 20.7 million rental units in the nation’s housing stock that were aﬀordable to these renter households. There were 9.9 million households with income below the 30% level; only 6.8 million units were aﬀordable to households with income below the 30% of AMFI.

The nationwide counts of renter households and rental units in each category can be compared to identify any mismatch between the supply and demand of aﬀordable rental housing. Renters in the 30% to 60% AMFI category are well served as these households had more units available than there were households (see Table 1). The surplus was of considerable scale, about 11 million units. This is not a trivial surplus; it means that that the ratio of units to households was in excess of 2 to 1.

Unfortunately, this surplus did not extend to the poorest category of renter households. For those with income below 30% of AMFI, there were 9.9 million renter households and only 6.8 million units to house. Thus, there was a shortage of 3.1 million units or, crudely, two housing units for every three households in this income category. This left about 31% of the households in this income category without aﬀordable housing.

The pattern of the mismatch appears to be consistent over time, as the pattern in 2005 is roughly the same. Note that direct comparisons of these two time periods are somewhat problematic as the 2000 count is from census data tabulated at the tract level using individual income limits for each metropolitan level. The 2005 data are national data using the nation’s median family income. With this caveat, there appears to be a persistent shortage of rental housing aﬀordable to the poorest renters, those below 30% of AMFI, and a surplus of housing for the next higher category, those between 30% and 60% of AMFI. Despite this pattern, the LIHTC is adding units aﬀordable to renter households in the 30% to 60% category where a surplus exists. 5Tracts have been criticized as too large as they are larger than common perceptions of neighborhoods (Coulton, Krobin, Chan, and Su 2001). They may also have too much variation within their boundaries making it hard to determine the true level of market need with a neighborhood (Goodman and Thibodeau 2003). Tracts have been criticized as too small as they may be smaller than the area containing a submarket of housing type that can deﬁne a household’s search area (Clapp and Wang 2006). However, Census tracts are used here as they approximate a city district within which a surplus, shortage, or balanced supply of rental housing in one price range can be identiﬁed. While the larger market area could be balanced, any imbalance at the tract level demonstrates that a surplus condition is being made worse with the addition of LIHTC units. This condition could be missed if the areal unit is larger. Similarly, a smaller areal unit could give a false indication of a surplus as the addition of a single LIHTC project could create the surplus condition.

160K. McClure LIHTC units are generally too expensive for the renter households with incomes below the 30% level unless they have additional subsidy, such as through vouchers.

Nationally, a renter household in 2000 with income placing it at the 30th percentile of median family income could aﬀord a gross rent of no more than about $390. The rents on LIHTC are not linked to the income of the household but to the income of a renter household at either 50 or 60% of the AMFI, depending upon the developer’s decision. While the rents of LIHTC units are not known with precision, the typical LIHTC development in 2000 had rent limits ranging from $700 to $900, depending upon unit size and which income limitation was selected. Knowing that the LIHTC program creates incentives for property owners to charge rents close to this maximum, the rents in LIHTC developments are considerably above what a household could aﬀord with income below the 30% level.

The LIHTC developments appear to be at a price range that is too high to serve those renter households with extremely low-income (less than 30% of AMFI), and the program may be adding to a stock that is in a surplus condition nationwide.

However, the program could still be generating useful additions to the stock. First, it could be directing units to those housing submarkets that, despite the surplus nationally, are suﬀering from a shortage locally. These submarkets may have more renter households in the 30% to 60% of AMFI than there are units serving them. To the extent that the LIHTC program directs units into such tracts, it is responding to a need. Second, it could be directing units to those housing submarkets that provide few or no aﬀordable housing options at all. These submarkets may have a very small number of units at aﬀordable price levels with or without any households in need of such units. Such tracts have few or no aﬀordable units or low-income households, thus there are few or no aﬀordable housing alternatives independent of the issue of latent demand existing in the tract. To the extent that the LIHTC program directs units into such tracts, it is providing aﬀordable housing opportunities where none may have existed before. This serves the goal of dispersal of the poor and, possibly, provides low-income households with access to schools, shopping, and employment opportunities not previously available.

Examination of the approximately 65,000 tracts across the nation ﬁnds that the typical tract has a surplus of 168 rental units with gross rents aﬀordable to households in the 30 to 60% range (see Table 2). The typical census tract has about Table 2. Housing market charatersitics for tracts by metropolitan location.

Average for all tractsLocation of tracts Central City Suburban Non-metropolitan Total Demographic characteristics of tracts Total population 3,902 4,763 3,986 4,300 Total households 1,496 1,747 1,522 1,612 Rental units 790 507 437 586 Renter households 737 474 397 545 Units and households between 30 and 60 percent of AMFI Rental units 433 280 209 316 Renter households 202 130 101 148 Units net households 231 150 109 168 Surplus as percent of stock 53% 54% 52% 53% Source: US Bureau of the Census, Census 2000. Housing Policy Debate161 1,600 households. About 545 are renters housed in about 586 rental units. The renter households are distributed across a wide range of income levels, but an average of 148 households are in the 30% to 60% of AMFI level and are served by 316 rental units aﬀordable to them.

This surplus condition varies little by whether the tract is located in a central city, suburb, or non-metropolitan area. The average surplus is greater in central cities where the rental stock is larger, but the housing stock in central cities is generally older and often in inferior condition relative to that in the suburbs. The non- metropolitan tracts, those in small towns and rural areas, have the lowest surplus but on a smaller base. Whether in metropolitan locations or non-metropolitan, these surpluses of units for renters in the 30% to 60% of AMFI are a very consistent 52% to 54% of the stock.

This surplus of units does not mean that renter households in the 30 to 60% of AMFI category will ﬁnd units readily available. Rather, these units are available to all renter households, and many renter households with higher levels of income choose to occupy units in this price range to lower their housing cost burden. This causes many renter households in the 30% to 60% of AMFI category to suﬀer a high housing cost burden despite a surplus of units in the price range that is aﬀordable to them (McClure 2005a).

The surplus is not present in all tracts. Rather, tracts range widely from a severe shortage within the tract as high as 700 units to a surplus of over 3000 units. Across the 65,000 tracts in the nation, only about 4900 (7.6%) have a shortage of rental units for households in the 30% to 60% of AMFI category. Of these, about 4700 have a shortage of fewer than 100 units. In 2000, only 258 tracts nationwide reported a shortage of over 100 units (see Table 3).

The scale of the shortage is interesting. The total shortage of units in those tracts where a shortage exists is about 139,000 units. The total annual production of the LIHTC is about 100,000 units. Thus, if the units were targeted to these tracts with unmet demand for such units, then this total shortage could be resolved in a very short time frame.

Do LIHTC units enter markets where there is a shortage of housing in 30–60% of AMFI income-rent range?

The study now examines the number of LIHTC units added over a ﬁve-year period from 2000 through 2004, the last year for which complete year data are available from the US Department of Housing and Urban Development. At issue is whether or not the program is directing units to those approximately 4,900 census tracts where a demonstrable shortage of units exists.

The mean number of units built in tracts with a shortage in excess of 100 units was 106 units. However, the LIHTC developments that were developed in these tracts managed to make entry into only 36 of the 258 tracts with a large shortage.

The mean number of LIHTC units built in tracts with a shortage of 0 to 100 units was 64 units. Again the market entry issue was a problem locating in only 411 of the almost 4700 tracts with a small shortage of these units (see Table 4). Given the inability to penetrate all markets with a need, the LIHTC program added only about 30,000 units to these tracts with a shortage over a ﬁve-year period falling well short of resolving the need for 138,000 units. Thus, where the LIHTC program managed to penetrate the market, it was capable of satisfying the latent demand, in some cases 162K. McClure fully. However, in many more tracts, the LIHTC program did not enter the market despite the need for aﬀordable housing of this type. This may be due to lack of available sites to construct new developments or lack of available buildings to purchase for rehabilitation and conversion into LIHTC projects. It may also be a Table 4. Count of LITC units developed 2000–2004 in tracts by category of and renters with income between 30% and 60% of AMFI.

Category of surplus (shortage) of aﬀordable unitsLIHTC units Percent of Mean units Tracts developed all tracts per tract Entered Shortage greater than 100 units 3,823 0.8% 106 36 Shortage less than 100 units 26,269 5.7% 64 411 Surplus less than 100 units 111,768 24.4% 31 3,614 Surplus is 100 to 199 units 104,255 22.7% 28 3,682 Surplus is 200 to 299 units 68,931 15.0% 30 2,277 Surplus is 300 to 399 units 48,156 10.5% 39 1,248 Surplus is 400 to 499 units 30,724 6.7% 42 734 Surplus is 500 to 599 units 19,649 4.3% 43 460 Surplus is 600 to 699 units 14,693 3.2% 50 296 Surplus is 700 to 799 units 9,450 2.1% 51 184 Surplus is 800 to 899 units 5,275 1.2% 45 118 Surplus is 900 to 999 units 4,149 0.9% 44 94 Surplus is 1000 to 1099 units 2,524 0.6% 46 55 Surplus is 1100 to 1199 units 2,134 0.5% 61 35 Surplus is 1200þunits 6,716 1.5% 64 105 Total 458,516 100% 34 13,349 Source: US Bureau of the Census, Census 2000.

US Dept. of HUD, LIHTC Database.

Table 3. Count of census tracts by category of diﬀerence between units aﬀordable to and renters with income between 30% and 60% of AMFI.

Category of surplus (shortage) of aﬀordable unitsAll Tracts in CategoryPercent of all tractsUnit Surplus (Shortage)QCTs in CategoryPercent of QCTsUnit Surplus (Shortage) Shortage greater than 100 units258 0.4%751,893 60 0.6%710,333 Shortage less than 100 units 4,688 7.2%787,029 605 6.1%715,842 Surplus less than 100 units 27,196 41.6% 1,200,906 2,965 29.8% 142,301 Surplus is 100 to 199 units 14,258 21.8% 2,069,037 2,329 23.4% 343,506 Surplus is 200 to 299 units 7,809 11.9% 1,912,367 1,493 15.0% 365,696 Surplus is 300 to 399 units 4,305 6.6% 1,486,215 885 8.9% 305,215 Surplus is 400 to 499 units 2,531 3.9% 1,127,193 566 5.7% 251,712 Surplus is 500 to 599 units 1,533 2.3% 835,882 369 3.7% 200,728 Surplus is 600 to 699 units 924 1.4% 595,934 208 2.1% 133,739 Surplus is 700 to 799 units 611 0.9% 456,099 149 1.5% 111,294 Surplus is 800 to 899 units 395 0.6% 334,936 89 0.9% 75,192 Surplus is 900 to 999 units 292 0.4% 276,364 83 0.8% 78,470 Surplus is 1000 to 1099 units 194 0.3% 203,112 46 0.5% 48,165 Surplus is 1100 to 1199 units 108 0.2% 123,792 26 0.3% 29,999 Surplus is 1200þunits 341 0.5% 516,911 80 0.8% 117,695 Total 65,443 100% 9,953 100% Source: US Bureau of the Census, Census 2000. Housing Policy Debate163 result of political opposition to the development of assisted housing despite a need for this type of housing.

There is an additional layer to this problem. In these ﬁve years, the LIHTC program added 458,000 units (for which locations are known), of these 428,000 went to tracts that already had a surplus of units. If anything, there was a slight tendency toward tracts with larger surpluses. There is a weak positive correlation between the surplus of units in each tract and the number of LIHTC units built from 2000 through 2004. (see Table 5). This suggests a tendency to build more LIHTC units in tracts where the surplus is greater rather than where a shortage exists. While the correlation is statistically signiﬁcant, the relationship is very weak. This tendency to develop units where there is a previously existing surplus may be driven by the lack of political opposition to the development of additional assisted housing. However, it suggests that the LIHTC program, despite the market analysis requirement, is largely indiﬀerent to market need in the placement of aﬀordable units; if anything, it leans toward providing units where they are not needed.

It is possible that this indiﬀerence to a need for units varies across metropolitan areas from the central cities, to the suburbs, as well as non-metropolitan areas with its small towns and rural districts. The variation could be the result of a bias toward developing units in older neighborhoods, especially those in central cities where the focus is more on redevelopment of the area rather than responding to a perceived need for rental housing aﬀordable to the target population. However, as Table 6 shows, it appears that the LIHTC program’s propensity to locate units into areas where a surplus of aﬀordable housing already exists is not a function of metropolitan location.

The central cities and the suburbs send comparable percentage of units to the tracts where there are shortages of units. The non-metropolitan areas send a lower percentage of tax credit units to these tracts, but this is on a much smaller base, reducing its signiﬁcance. All three areas send over 90% of their LIHTC units to those areas with a surplus. While there are some minor variations, there is no clear pattern showing a greater responsiveness to need in any one metropolitan location.

It is also possible that the location of LIHTC units into tracts with a surplus is the result of the additional subsidy given to development located in QCTs. Of the total of about 65,000 tracts nationwide, about 10,000 are QCTs by virtue of their high incidence of low-income households. These tracts do not necessarily have a Table 5. Correlation between the count of LIHTC units developed 2000–2004 in each tract and the surplus (shortage) of units for renter households with income between 30% and 60% of AMFI.

Correlation test Variables: Count of rental units in tract aﬀordable to households 30-60% AMFI minus renter households 30–60% AMFI LIHTC units added to tract during the years 2000 through 2004 Pearson Correlation 0.060 Signiﬁcance (one-tailed) 0.00001 Number of tracts: 13,349 Source: US Bureau of the Census, Census 2000.

US Dept. of HUD, LIHTC Database.

164K. McClure Table 6. Count of LIHTC units developed by metropolitan location and level of market need.

LIHTC units developed 2000 to 2004 Metropolitan Location Census tracts categorized by renter housheolds with income between 30% and 60% of AMFI minusCentral City Suburbs Non-Metro Total rental units aﬀordable to these households Units Units Units Units All Tracts Tracts with a shortage 17,616 8.4% 10,841 5.7% 1,635 2.8% 30,092 6.6% Tracts with a surplus of 0–200 units 88,027 42.0% 92,754 48.8% 35,242 59.8% 216,023 47.1% Tracts with a surplus of 200–800 units 93,609 44.7% 75,943 39.9% 22,051 37.4% 191,603 41.8% Tracts with a surplus of 800þunits 10,116 4.8% 10,682 5.6% – 0.0% 20,798 4.5% All tracts 209,368 100% 190,220 100% 58,928 100% 458,516 100% Qualﬁed Census QCTs with a shortage 13,627 11.7% 680 2.7% 56 0.6% 14,363 9.5% Tracts QCTs with a surplus of 0–200 units 50,208 43.2% 11,795 47.4% 5,910 58.9% 67,913 45.0% QCTs with a surplus of 200–800 units 47,869 41.2% 10,539 42.4% 4,065 40.5% 62,473 41.4% QCTs with a surplus of 800þunits 4,478 3.9% 1,856 7.5% – 0.0% 6,334 4.2% All Qualiﬁed Census Tracts 116,182 100% 24,870 100% 10,031 100% 151,083 100% Percent of Units New ConstructionPercent of Units New ConstructionPercent of Units New ConstructionPercent of Units New Construction New Construction Tracts with a shortage 42.2% 85.2% 94.4% 62.4% Tracts with a surplus of 0–200 units 55.9% 76.1% 69.3% 66.8% Tracts with a surplus of 200–800 units 49.3% 67.3% 64.0% 58.2% Tracts with a surplus of 800þunits 56.7% 44.2% 0.0% 50.2% All tracts 51.9% 71.2% 68.0% 62.1% Housing Policy Debate165 shortage of units aﬀordable to households with income between 30% and 60% of AMFI. Table 6 shows that the level of shortage of aﬀordable units among QCTs is very similar to all tracts. About 7.6% of all tracts have a shortage as do the same percentage of the QCTs. However, LIHTC units are under-represented among all tracts with a shortage, but they are over-represented in QCTs.

The proportion of LIHTC units in QCTs is too small to oﬀset the overall tendency of LIHTC units to be located in tracts that are in surplus. The subsidy bonus given to LIHTC developments in QCTs may cause a modest shift of LIHTC units toward tracts with a shortage than might be expected. Only about 14,400 units were built in QCTs with a shortage of aﬀordable units. This is 9.5% of the units built in QCTs. If the designation of the tract as a QCT had no eﬀect on location, then the allocation to tracts with a shortage would be only about 5.1% of the total (the share of non-QCT tracts in areas with a shortage of aﬀordable units). This means that about 6,600 more units were built in areas with a shortage of aﬀordable housing because of the QCT bonus. This is rather small in comparison to the over 458,000 LIHTC units built over the ﬁve years study period. However, it does suggest that the boost in subsidy provided through the QCT process can have a positive inﬂuence on the location of investments. Possibly this positive inﬂuence could be better directed if it was tied to the presence of market need (demand in excess of supply) rather than the presence of poverty independent of the scale of the supply of aﬀordable housing.

It is also possible that the tendency for LIHTC units to be located in housing markets with a surplus can be explained by construction type. LIHTC units can be developed either through the rehabilitation of an existing unit or the construction of a new unit. The data identify the construction type of all LIHTC developments; unfortunately the data do not identify the use of the site prior to the development. A new construction project may have been built on a site where an existing building once stood. Thus, the newly constructed units may not indicate a net gain in housing units on the site as existing units may have been demolished prior to construction of the LIHTC units. Equally, the presence of a rehabilitation project does not mean that no units were added to the stock. The rehabilitated units may have been developed out of non-housing space generating a net gain in units. However, it can be reasonably assumed that new construction units represent the addition of housing units to tracts and that rehabilitated units represent a much smaller addition to the tracts. If the new construction units are disproportionately locating in tracts with a shortage of aﬀordable housing and rehabilitated units are locating in tracts with a surplus, then the LIHTC program may be more eﬀective than it appears.

Table 6 lists the percentage of new construction units found in each market type.

The expectation is that new construction should dominate in the tracts with a shortage, but this is not found. Overall, about 62.1% of LIHTC units built from 2000 through 2004 were newly constructed, and a similar 62.4% were newly constructed in tracts with a shortage of units aﬀordable to households in the 30% to 60% of AMFI. There was variation across metropolitan areas. In the suburban and non-metropolitan tracks with a shortage of aﬀordable units, the LIHTC did favor new construction as expected. In the central cities, it did not. Rehabilitation would be expected to dominate in the tracts with a surplus, but the results were mixed. In tracts with a large surplus of aﬀordable units of 200 or more units, there was a small but measureable shift toward rehabilitation. However, even this shift of only a few percentage points was not consistent across the metropolitan areas. In the tracts with a small surplus of units, between 0 and 200 units, there was actually a bias in favor of 166K. McClure Table 7. Entry of LIHTC units into tracts with few aﬀordable units by metropolitan location and share of rental stock.

Percent of rental units aﬀordable to householdMetropolitan Location in the 30 to 60 percent of AMFI Central City Suburbs Non-Metro Total Count of LIHTC units developed 2000 to 2004 Less than 20% of stock 2,858 1.4% 7,168 3.8% 901 1.5% 10,927 2.4% Between 20% and 70% of stock 36,834 17.6% 67,071 35.3% 22,439 38.1% 126,344 27.6% Greater than 70% of stock 169,363 81.0% 115,733 60.9% 35,588 60.4% 320,684 70.0% Total 209,055 100% 189,972 100% 58,928 100% 457,955 100% Percent of the rental stock Less than 20% of stock 2.5% 4.6% 1.1% 3.1% Between 20% and 70% of stock 27.1% 35.9% 38.6% 32.4% Greater than 70% of stock 70.4% 59.5% 60.3% 64.5% Total 100% 100% 100% 100% Diﬀerence in percentage of rental stock and percentage of LIHTC untis developd 2000 to 2004 Less than 20% of stock71.170.8 0.470.7 Between 20% and 70% of stock79.570.670.574.8 Greater than 70% of stock 10.6 1.4 0.1 5.5 Source: US Bureau of the Census, Census 2000.

US Dept. of HUD, LIHTC Database. Housing Policy Debate167 new construction. The distribution of LIHTC units between new construction, which adds units to the stock, and rehabilitation, which renovates existing units, appears to explain only a very small part of the placement of units into tracts with an existing surplus of aﬀordable units. This explanation is weak in that well over one-half of all LIHTC units added to markets in surplus are newly constructed suggesting that these LIHTC units are expanding already saturated markets.

Do LIHTC units enter markets with few aﬀordable housing units?

Among the 65,000 tracts in the nation, on average over two-thirds of the rental units are aﬀordable to renter households with income below 60% of AMFI. However, the distribution is not normal; it is skewed to the right with a large percentage of tracts having an above average percentage of these aﬀordable units. About 57% of all tracts have more than 70% of their rental units aﬀordable to households with income at 60% of AMFI.

At the other extreme, only 5% of all tracts have less than 20% of their rental stock priced in the aﬀordable range. These tracts tend to have fewer rental units in total compared to the average tract. The average rent in these tracts is 49% higher than the average tract. To the extent that these tracts have renters with modest income, there is a shortage of aﬀordable units relative to the counts of these households. Typically these tracts have a modest shortage of 23 units compared to the average surplus of 170 units found in all tracts. Thus, these tracts could be thought of as desirable places to develop LIHTC units as few, if any, aﬀordable housing options have been oﬀered there.

Table 7 illustrates that the LIHTC program is not making entry into those tracts that have a very low percentage of aﬀordable units. Tracts with less than 20% aﬀordable rental housing contain only 3.1% of the rental stock, but the LIHTC program placed only 2.4% of recently developed units in these tracts. This failure to achieve market entry for purposes of providing aﬀordable housing opportunities where they have not previously existed is especially true for the central city tracts.

Central city tracts with little aﬀordable housing contain 2.5% of the rental housing stock but attracted only 1.4% of the recently developed LIHTC units, a drop of 1.1 percentage points. In suburban tracts of this type, the drop was marginally less with 0.8 percentage points. Only in the non-metropolitan areas did the LIHTC program place units in the tracts on a par with the percentage of the rental stock.

At the other extreme in tracts with an above average percentage of aﬀordable units, the LIHTC program tended to place more units in these tracts than would be expected. About 65% of all rental units are in such tracts, but 70% of recently developed LIHTC units located there. This placement of LIHTC units into areas with an already high concentration of aﬀordable housing is especially strong in central cities.

Conclusion There is a clear need for additional aﬀordable rental housing in the United States.

There is a persistent shortage of rental units aﬀordable to the poorest among the renter households, but the shortage is in the lowest price range, units aﬀordable to renter households with income below 30% of AMFI. There is a surplus of units in the next higher category of renter households, those with income between 30% and 168K. McClure 60% of AMFI. The LIHTC program is adding to this sector of the rental housing stock, a stock that is in surplus.

Using scarce public resources to add to a stock that is in surplus could be justiﬁable if the LIHTC program is directing units to just those tracts that have a shortage of units aﬀordable to households with income in the 30% to 60% of AMFI.

This analysis ﬁnds that the program is not directing units where shortages exist and that this is true across all locations, whether in central cities, suburbs, or non- metropolitan areas.

Using scarce public resources to add to a stock that is in surplus could be justiﬁable if the LIHTC program is directing units to those tracts that provide few aﬀordable housing options, independent of any shortage of aﬀordable units relative to the count of low-income households. If the program provided units in these markets, it could be promoting economic integration and dispersing the poor, both of which are laudable goals. This analysis ﬁnds that the program is not directing units where few or no aﬀordable housing options exist. In fact, it appears to be doing the opposite; it is concentrating units where an above average percentage of these units already exists. This furthers the segregation of the population by income level, increasing the social isolation of the poor.

These results suggest that the implementation strategy for the LIHTC program is not working very well. The tax credits are allocated to developers through competitions held by state housing ﬁnance agencies. These agencies allocate these units based upon a QAP. While this plan is federally mandated, states are granted wide latitude to use the federal tax credits in the manner best suited to the needs identiﬁed by each state. The analysis here suggests that the QAPs do not seem to direct these scarce federal housing resources to areas where there is a demonstrated need for such housing or where few aﬀordable housing options exist independent of need.

The analysis also suggests that the implementation of the LIHTC program is ﬂawed in another way. The program requires the state agencies that allocate the tax credits to require from each developer a market analysis that identiﬁes that there is a latent demand for such housing. The results presented here indicate that the market analyses are less than accurate. Overwhelmingly, the LIHTC units have been developed where aﬀordable units are already in surplus.

Perhaps the government should re-examine the implementation of the LIHTC program. It may be time to recognize that not all markets need added units. Many markets need resources to help extremely low-income renter households rent units that are already in the market, but are priced out of their reach. Katz and Turner (2008) have called for revising the system through which the tax credits are allocated.

They suggest that credits should be directed to markets where there is a demonstrable need for production of units. Many markets need resources to renovate existing units, which is possible within the LIHTC program. However, this means that the QAPs need to direct the developers into this approach rather than adding units to submarkets that already have a glut of units.

References Abravanel, Martin D., and Jennifer E. H. Johnson. 2000.The low-income housing tax credit program: A national survey of property owners. Washington, DC: US Department of Housing and Urban Development, Oﬃce of Policy Development and Research. Housing Policy Debate169 Burman, Leonard E. 2008.Tax topics: Low-income housing credit. Washington, DC:

The Urban Institute and Brookings Institution Tax Policy Center. World Wide Web page5http://www.taxpolitycenter.org/taxtioucs/encycolopedia/Low-Income-Housing.

cfm4(accessed September 15, 2008).

Buron, Larry, Sandra Nolden, Kathleen Heintz, and Julie Steward. 2000.Assessment of the economic and social characteristics of LIHTC residents and neighborhoods. Cambridge, MA: Abt Associates for the US Department of Housing and Urban Development, Oﬃce of Policy Development and Research.

Case, Karl E. 1991. Investors, developers, and supply-side subsidies: How much is enough?

Housing Policy Debate2, no. 2: 341–56.

Clapp, John M., and Yazhen Wang. 2006. Deﬁning neighborhood boundaries: Are census tracts obsolete?Journal of Urban Economics59, no. 2: 259–84.

Coulton, Claudia J., Jill Korbin, Tsui Chan, and Marilyn Su. 2001. Mapping residents’ perceptions of neighborhood boundaries: A methodological note.American Journal of Community Psychology29, no. 2: 371–83.

Cummings, Jean L., and Denise DiPasquale. 1999. The low-income housing tax credit:

Analysis of the ﬁrst ten years.Housing Policy Debate10, no. 2: 251–307.

Downs, Anthony. 1981.Neighborhoods and urban development. Washington, DC: The Brookings Institution.

Ernst & Young LLP, Kenneth Leventhal Real Estate Group. 1997.The low-income housing tax credit: The ﬁrst decade. Boston: Ernst & Young LLP.

Freeman, Lance. 2004. Siting aﬀordable housing: Location and neighborhood trends of low- income housing tax credit developments in the 1990s. Washington, DC: Brookings Center on Urban and Metropolitan Policy, Census 2000 Survey Series. World Wide Web page 5http://www.brookings.edu/urban/pubs/20040405_Freeman.pdf4(accessed November 14, 2005).

Freeman, Lance. 2006. Comment on Kirk McClure’s ‘‘The low-income housing tax credit programs goes mainstream and moves to the suburbs.’’Housing Policy Debate17, no. 3:

447–59.

Goodman, Allen C., and Thomas G. Thibodeau. 2002. Housing market segmentation and hedonic prediction accuracy.Journal of Housing Economics12, no. 3: 181–201.

Guggenheim, Joseph. 2003.Tax credits for low income housing: Opportunities for developers, non-proﬁts, agencies and communities under expanded tax code provisions, 12th ed. Glen Echo, ML: Simon.

Haley, Barbara A. 2008. Guest editor’s introduction.Cityscape10, no. 1: 1–4.

Katz, Bruce, and Margery Austin Turner. 2008. A new blueprint for federal, state and local action. InRevisiting rental housing: policies, programs, and priorities, eds. Nicholas P.

Retsinas, and Eric S. Belsky, 319–58. Cambridge, MA: Joint Center for Housing Studies, Harvard University and Washington, DC: Brookings Institution Press.

Knaap, Gerrit. 1998. The determinants of residential property values: Implications for metropolitan planning.Journal of Planning Literature12, no. 3: 267–82.

Malpezzi, Stephen, and Kerry Vandell. 2002. Does the low-income housing tax credit increase the supply of housing?Journal of Housing Economics11, no. 4: 360–80.

McClure, Kirk. 2000. The low-income housing tax credit program as an aid to housing ﬁnance: How well has it worked?Housing Policy Debate11, no. 1: 91–114.

McClure, Kirk. 2005a. Deciphering the need in housing markets: A technique to identify appropriate housing strategies at the local level.Journal of Planning Education and Research24, no. 4: 361–78.

McClure, Kirk. 2005b. Rent burden in the housing choice voucher program.Cityscape: A Journal of Policy Development and Research8, no. 2: 5–20.

McClure, Kirk. 2006. The low-income housing tax credit program goes mainstream and moves to the suburbs.Housing Policy Debate17, no. 3: 419–46.

Nelson, Kathryn P. 1994. Whose shortage of aﬀordable housing?Housing Policy Debate5, no.

4: 401–42.

Oakley, Deirdre. 2008. Locational patterns of low-income housing tax credit developments: A sociospatial analysis of four metropolitan areas.Urban Aﬀairs Review 43, no. 5: 599–628.

Stegman, Michael A. 1991. The excessive cost of creative ﬁnance: Growing ineﬃciencies in the production of low-income housing.Housing Policy Debate2, no. 2: 357–73. 170K. McClure US Census Bureau. 2005. American housing survey. Retrieved February 11, 2008 from http:// www.census.gov/hhes/www/housing/ahs/ahs.html.

US Department of Housing and Urban Development. 2008a. Housing choice vouchers fact sheet. Retrieved September 18, 2008 from http://www.hud.gov/oﬃces/pih/programs/hcv/ about/fact_sheet.cfm.

US Department of Housing and Urban Development. 2008b. Low-income housing tax credits database. Retrieved February 11, 2008 from http://www.huduser.org/datasets/lihtc.html.

US Government Accounting Oﬃce. 1997. Opportunities to improve oversight of the low- income housing program. GAO/GGD/RCED-97-55 Low Income Housing Tax Credit.

Washington, DC: US Government Accounting Oﬃce. Housing Policy Debate171