Data Examined from Several Sources

Most information about LIHTC activity traditionally has come from the National Council of State Housing Agencies (NCSHA), the nonprofit organization whose members are state Housing Finance Agencies. NCSHA’s data includes the total allocations of credits at the time they are made, but it does not provide information about the lag time between credit allocations and the time projects are placed in service. It also does not provide characteristics of individual properties, because NCSHA collects data at a very broad aggregate level (entire states).

The General Accountability Office (GAO) collected information on LIHTC projects placed in service between 1992 and 1994.¹  City Research of Chicago collected more detailed information from a sample of projects placed in service between 1987 and 1996.²   Both efforts were one-time only research projects, where the data were collected once and never subsequently updated.

A more valuable database is maintained by HUD and its contractor Abt Associates Inc. because that database has been regularly updated since it was first compiled in 1997.  HUD’s LIHTC Data Base is compiled by collecting data from the HFAsand its most recent update includes projects placed in service through 2004.

HUD released its most recent update of theLIHTC Data Base  last month, along with a report that tabulated many of the basic characteristics of these properties.  (Readers interested in general statistics are encouraged to consult that report.³ )  

The HUD LIHTC Data Base contains information on roughly 24,500 LIHTC projects with 1.2 million rent-restricted units placed in service through 2004.⁴ NCSHA’s State HFA FactBook shows cumulative allocations for about 1.5 million rent-restricted units (including tax-exempt bond allocations) through 2003. Given lags and other differences between allocations and placing units in service, as well as gaps in the data reported by HFAs (despite follow-ups, some projects in the HUD LIHTC Data Base don’t report number of units, for example), the difference between HUD- and NCSHA-reported totals is not excessive.

The NCSHA and HUD numbers for credits allocated between 1998 and 2002 are compared in Table 1 (above). Data are available in a consistent format from NCSHA beginning in 1998, and a large share of credits allocated after 2002 would not have been placed in service yet by 2004 (the cut-off year in the HUD LIHTC Data Base). 

Allocations: Nonprofit vs. For-profit

Over that time horizon, dollars allocated to nonprofits have remained relatively close to 30% of total (non-tax-exempt bond) allocations. Nonprofits also accounted for roughly 30% of the new construction LIHTC projects in 1998 and 1999, but the percentage dropped afterwards—all the way to 22% in 2002. 

A caveat: There is possible ambiguity over the definition of nonprofit developer, as some LIHTC developers classified as for-profit could include joint ventures with nonprofit partners.  Little can be done to adjust the data for this possibility, so readers are advised to keep this in mind when viewing the following tables and graphs.

Because for-profit developers tend to build larger projects,⁵ the nonprofit share is generally smaller if the percentage is based on the number of units, rather than projects-placed-in-service.  However, the trend toward a lower non-profit share is the same.

Thus, Table 1 shows an apparent discrepancy, in that NCSHA’s nonprofit share of allocations remains relatively constant, while the HUD numbers show a strong decline, especially in the most recent year. The universe for each of the two sets of numbers is different, however. The NCSHA statistics include rehab projects, but exclude projects with tax-exempt bonds. The statistics based on the HUD LIHTC Data Base exclude rehab projects but include those with tax-exempt bonds.

If, for some reason, nonprofit projects experience a greater lag between the time they receive LIHTC allocations and the time they’re placed in service, this could explain some of the decline in the nonprofit shares in Table 1, particularly for the most recent allocation year shown, 2002. An apparent decline would occur if a disproportionate share of nonprofit units receiving allocations in 2002 were still in the queue waiting to be placed in service in recent years. This can’t be tested directly for the year 2002 with current data, but in earlier years the share of projects with a two-or-more year lag between the allocation and placed-in-service dates was higher for nonprofits than  it was for for-profit developers—although the gap has been narrowing and disappeared entirely in 2001 (Figure 1). 
 

Key Differences between Nonprofits and For-Profits

An assumption underlying the nonprofit set-aside is that nonprofits behave differently than for-profit developers do. The HUD LIHTC Data Base can be used to look at some of the differences, and how they have changed over time.

A key change in the LIHTC program over the past decade is the increase in the use of 4% credits in combination with tax-exempt bond financing. The increase has occurred primarily among for-profit developers, for whom the share of LIHTC projects placed in service with tax-exempt bonds increases from roughly 3% to 30% between 1995 and 2001 (Figure 2).

Over the same time, the share of projects with 4% credits (indicating that they are receiving some type of federal subsidy) but without tax-exempt bonds has declined (Figure 3). 

Tax-exempt bonds differ from other federal subsidies that reduce the credit percentage from 9% to 4% in that they are not subject to the per capita limit on LIHTCs (under the theory that the total amount of private-activity, tax-exempt bond authority is subject to its own per capita limitation).⁶ Other federal subsidies reduce the credit percentage without conveying this particular advantage.

More specific behavioral tendencies sometimes attributed to nonprofits include possibly stronger inclinations than for-profits to serve particularly distressed locations or particularly disadvantaged households. 

For example, it may be tempting to assume that nonprofits would be more likely to build LIHTC projects in central cities or non-metropolitan locations, while the for-profits would be more likely to focus on suburban areas where incomes are generally highest. However, over the 1995-2003 period, the share of for-profit projects placed in service in central cities has been slightly higher than the nonprofit share every year (Figure 4).

The share of for-profit projects placed in service in non-metropolitan areas has been somewhat higher than the nonprofit share in all but two of the years, when the for-profit and nonprofit shares have been about the same (Figure 5).

Two categories of distressed areas are called Qualified Census Tracts (QCTs) and Difficult Development Areas (DDAs). HUD maintains the lists of QCTs and DDAs, and LIHTC properties built in these areas are eligible for a 30% increase in basis, and therefore 30% more tax credits per dollar of project cost. For new LIHTC construction project placed in service over the 1995-2003 period, the share of nonprofit projects receiving an increased eligible basis has often, but not always, been greater than the for-profit share (Figure 6).

In terms of serving populations with special needs, a little more than 30% of the projects placed in service between 1995 and 2003 have targeted the elderly. In some years the elderly share has been higher among the nonprofits. In other years it has been higher among the for-profits. There seems to be no systematic tendency one way or the other (Figure 7).

Also of interest is the tendency of for-profit and nonprofit developers to target the disabled. But because the data are so incomplete (missing for more than 21,000 properties), it’s impossible to say much about this based on the information in HUD’s LIHTC Data Base.

Are For-Profits More Efficient?

Questions about costs can’t be investigated with recent data, since neither NCSHA’s published reports nor HUD’s LIHTC Database contain the relevant information. Both GAO and City Research collected cost information in their studies, however, and reports from both organizations included results on cost differences between for-profit and non profit developers. Although somewhat dated, these reports still provide the best information available on the subject, so a brief review is worthwhile here.

First, there is a solid economic reason to suspect that for-profit developers create housing at lower cost. In a competitive industry, the profit motive forces businesses to operate more efficiently and learn how to control production costs. Nonprofit firms, not subject to these competitive pressures, may operate less efficiently and experience higher production costs.

According to the City Research study, tax credit projects with non-profit sponsors have development costs that are 43% higher per unit. Although some factors (apartment size, type of construction, and neighborhood characteristics) explain part of the difference, even after statistically controlling for these factors, tax credit units cost 15% more if they are built by nonprofit developers.  

According to the two GAO reports, tax credit projects with non-profit sponsors have development costs that are on average $18,000 higher per unit. Again, there are some factors (type of area built in, type of structure, type of tenant the projects intend to serve) that partially explain the difference. But even after controlling for those factors, the nonprofits’ development costs are still $5,600 higher per unit.

GAO’s statistical analysis of this number shows a better than 85% chance that for-profit developers build LIHTC units for lower cost than nonprofits. Because that’s slightly below the 90% or 95% often used as rule of thumb, the Executive Summary classifies the result as statistically insignificant. However, the GAO result doesn’t miss the 90% threshold by much. Moreover, GAO’s best point estimate is still that, even after controlling for a large number of other factors, a for-profit developer can build an LIHTC unit for $5,600 less than a nonprofit developer can.

----------------------------

  ^1. U.S. General Accounting Office, Tax Credits: Opportunities to Improve Oversight of the Low-Income Housing Program, GAO/GGD/RCED-97-55, March 1997; and U.S. General Accounting Office, Tax Credits: Reasons for Cost Differences in Housing Built by For-Profit and Nonprofit Developers, GAO/RECED-99-60, March 1999.
 ^2. Jean L. Cummings and Denise DiPasquale, Notes on Building Affordable Rental Housing: An Analysis of the Low-Income Housing Tax Credit, City Research, February 1998.
^3. HUD National Low Income Housing Tax Credit (LIHTC) Database: Projects Placed in Service through 2004; http://www.huduser.org/Datasets/lihtc/tables9504.pdfhttp://www.huduser.org/datasets/lihtc.html#about
^4. http://www.huduser.org/datasets/lihtc.html#about
^5. According to HUD’s analysis of projects placed in service 1995-2004, average project size is 72 units, but only 54 for nonprofit sponsors.
^6. Tax credit allocations are limited to 70 percent present value of a project’s eligible costs unless the project receives another type of federal subsidy such as tax-exempt bonds, in which case the limit is 30 percent of the present value.  At interest rates that have prevailed since the LIHTC program began, the present value calculations work out to approximately 9 percent and 4 percent of the cost annually.  The common practice is to refer to these as “9 percent” and “4 percent” credits.

  [ return to top ]