The quest to raise homeownership rates. (Cover Report: Industry Trends).

Studies of the most effective ways to raise homeownership rates suggest direct cash assistance gets the most bang for the buck. The simulations of home-buying behavior show cash assistance outstrips the effectiveness of the lower mortgage rates brought about by conforming loans.

A POLICY PRIORITY IN THE UNITED STATES IS to increase the rate of homeownership. To achieve that objective, policy-makers rely on a host of policies and programs that reallocate billions of dollars of resources. Several of these policies and programs try to increase homeownership by reducing mortgage rates. More specifically, federal sponsorship for Fannie Mae and Freddie Mac is one of the major tools that policymakers rely on to reduce mortgage rates.

Given the public resources involved, it seems reasonable for the general public, mortgage and housing professionals, and policymakers to consider the degree to which the mortgage rate reduction produced by Fannie Mae and Freddie Mac increases homeownership. Most of the evidence reviewed in this article finds that mortgage rate changes need to be around 2 percentage points before they have what many would consider a modest but not trivial effect on homeownership. Because Fannie Mae and Freddie Mac reduce mortgage rates on the order of 20 to 50 basis points, the effect of their rate reduction on homeownership is likely to be quite modest although, again, not trivial.

Moreover, the evidence presented here also suggests that a more direct method of subsidizing potential homeowners would have a larger effect on homeownership (while using the same amount of resources) than the reductions in mortgage rates attributed to Fannie Mae and Freddie Mac.

Several caveats are in order when considering these conclusions. Analysis of homeownership and mortgage rates is complicated by a number of factors, including the complexity of the decision to own and data concerns. As a result, the studies summarized all have important weaknesses. In addition, Fannie Mae and Freddie Mac could increase homeownership through means other than lowering rates, such as encouraging reductions in down payments.

Moreover, the two firms have several explicit legislative objectives, and "increasing homeownership" is not one of them. Even with these limitations, the evidence discussed here appears relevant for the existing, vigorous public discussion centered on mortgage rates, homeownership and Fannie and Freddie's activities.

Mortgage rates and homeownership

The 2002 annual report of the Federal Reserve Bank of Minneapolis features a discussion of how changes in mortgage rates affect homeownership (available online at www. minneapolisfed.org). I have summarized the material from that discussion in this article, focusing on simulation evidence. Specifically, analysts use two types of simulations to examine how changes in mortgage rates affect homeownership. In the first type, they simulate the loan underwriting process to determine how mortgage rates affect the ability of households to qualify for a mortgage.

In this underwriting simulation approach, analysts choose a reference house. They then review financial data to determine the percentage of households or families that would qualify for a mortgage on the reference house using specified mortgage qualification criteria and a prevailing mortgage rate. The analysts can then adjust the qualification criteria and the mortgage rate to examine how the change alters the number of households or families that can qualify for a mortgage on the reference house.

Examples of this approach include "Who Could Afford to Buy a House in 1995," by Howard Savage, published by the U.S. Census Bureau in 1999, and "The Potential and Limitations of Mortgage Innovation in Fostering Homeownership in the United States," by David Listokin and colleagues, published in the Fannie Mae Foundation's Housing Policy Debate in 2001.

A problem with the underwriting simulations just discussed is they do not account for the likelihood that a household will actually want to buy a house. For example, some renters may not want to own a home even if they could qualify for a loan. To address that limitation, a number of analysts take a second "tenure-choice" approach in which they model the probability of a household owning a home as a function of several factors (which I greatly simplify in this article).

In a recent example, the variables used in such modeling include estimates of the relative price of housing; an estimate of the permanent income of the household; demographic variables such as household size, age, race and gender; and an estimate of whether a household was prevented, or constrained, from buying a desired house because it could not meet a variety of underwriting guidelines.

After estimating all of the necessary variables, the authors calculate the probability of ownership using various down payment requirements, housing debt-to-income requirements and mortgage rates. They then compare the probability of ownership resulting from the various scenarios. The comparisons indicate how changes in mortgage standards and mortgage rates affect the probability of ownership.

Examples of this approach can be found in "The Impacts of Affordable Lending Efforts on Homeownership Rates," by Roberto Quercia and colleagues for Freddie Mac in 2000, and "Implications of Privatization: The Attainment of Social Goals," by Susan Wachter and colleagues for several government agencies in 1996.

Figure 1 summarizes the results of six simulations conducted by many of the analysts already mentioned and using both simulation methods (the extended version of this article provides important details about differences and similarities between the simulations and discusses some additional analysis).

Three of the simulations examined a reduction in mortgage rates of 2 percentage points while a fourth examined a reduction of 3 percentage points. These reductions had virtually no effect on homeownership for African-American households, and would increase homeownership for all households by roughly 50 basis points (the range was from 80 basis points to -10 basis points).

The other two simulations examine mortgage rate increases, one of 2 percentage points and one of 50 basis points. The larger increase had virtually no effect on homeownership. The smaller mortgage rate increase had the largest effect on homeownership of all the simulations, reducing homeownership by roughly 3 percentage points for African-American households and around i percentage point for all households. However, when updating this last simulation, the authors noted that a technical weakness in methodology had led it to overestimate effects on homeownership.

By way of context, 1 percent of renting households in 2000 equaled roughly 360,000. The average annual change in homeownership rates from 1960 to 2001 is 20 basis points. The average annual change in homeownership rates from 1995 to 2001 is 80 basis points.

What explains the central tendency of these results, which some might find smaller than expected? Simply put, changes in mortgage rates may not address all the barriers to homeownership for many households.

A lower mortgage rate allows more households to meet debt-to-income standards and could induce a household to decide to buy. But a number of factors beyond mortgage qualification standards influence the ownership decision, including income, the relative price of ownership and demographic factors such as age and family structure. And, most important, many households in the simulations also appear to have insufficient cash to pay a standard down payment and closing costs.

According to the simulations conducted by the Census Bureau, renting households typically have both wealth and income constraints. In the Census sample, 70 percent of renters have an inability to pay a down payment and closing costs, and too little income to meet debt-service requirements. Only 2 percent of renters are constrained by income alone.

The authors of the simulations conducted for Freddie Mac also found that limited wealth prevents lower rates from having a large effect on homeownership.

The authors note, "Consistent with the literature, the down payment requirement is a greater detriment to home purchase than the income requirement. Thus, lowering the cost of borrowing does not necessarily allow more people to purchase, once the down payment requirement becomes binding. For instance, although the percentage of income-constrained households decreases as a result of a 200 basis point drop from 8 percent to 6 percent in the interest rate, the percent of people that could actually buy a house remained the same because the percentage of down payment--constrained households remained unchanged. This implies that there is a significant overlap between the two constrained measures. Because lack of wealth to meet the necessary down payment is the dominant constraint, most households that are income-constrained are also wealth-constrained. However, the reverse is not the case.

The issue of credit quality

Of course, mortgage lenders also consider credit quality when making loans. The borrowers may not qualify for a loan because of their credit quality or, perhaps more likely, they will not meet the higher down payment or mortgage rate the lender requires to compensate for low credit quality. It is important to note, therefore, that none of the simulations considers credit quality. As a result, some of the households that qualify for a mortgage with the lower rate in the simulation may not actually qualify because of a low credit score.

Suggesting the potential importance of credit quality, research by Stuart Rosenthal for the Research Institute for Housing America (RIHA) found that removal of credit constraints could increase the home-ownership rate by as much as 4 percentage points. A paper prepared for the Millennial Housing Commission by Michael Collins analyzed Home Mortgage Disclosure Act (HMDA) data and found that poor credit history is the most frequently cited reason by mortgage originators for the denial of single-family mortgages.

The simulations, not surprisingly, have other weaknesses. Some of the data used in the analysis comes from surveys. Survey respondents may not correctly report the value of their assets or their incomes. In addition, the simulations reflect how mortgage rate changes or mortgage qualification standards affect households at a point in time. An increase in mortgage rates or a down payment requirement may delay, but not prevent, a household from owning a home.

The level and change in mortgage rates used in the simulations also seem to affect their results. Nonetheless, such simulations seem to be among the best available information on how mortgage rates affect homeownership. (Another analysis by professors Painter and Redfearn of the University of Southern California, using a completely different methodology, also found little effect on homeownership from changes in mortgage rates.)

With this information as background, I can provide some context to the mortgage rate changes credited to Fannie Mae and Freddie Mac.

Mortgage rate reductions by GSEs

One aspect of federal policy to increase homeownership rates is to reduce mortgage rates through interventions in secondary mortgage market activity. The federal government uses two distinct types of institutions active in secondary mortgage markets to lower mortgage rates. I focus on the government-sponsored enterprises (GSEs), Fannie Mae and Freddie Mac, because of the greater scope of their activities. Fannie and Freddie have financed more mortgages than Ginnie Mae, and guarantee both full and timely repayment of funds to investors.

As is well known, Fannie Mae and Freddie Mac are privately owned, publicly traded firms. At the same time, observers see them as "sponsored" by the federal government because of their public attributes, and this sponsorship reduces their costs. Perhaps most important, the implied guarantee on Fannie Mae and Freddie Mac debt that results from sponsorship reduces their borrowing costs by making securities safer and more liquid.

Because they have lower costs of raising funds, the GSEs can pay a higher price for mortgages than non-GSE competitors, thereby reducing the interest rate on mortgages while still earning sufficient returns to attract capital. In this way, the lower cost made possible by federal sponsorship can work its way into lower mortgage rates for households.

To estimate the degree to which the GSEs lower mortgage rates, analysts examine the difference in rates between conforming mortgages (those eligible for purchase by the GSEs) and jumbo mortgages, while trying to hold other factors constant. According to the most recent summaries of the research, analysts estimate that Fannie Mae and Freddie Mac lower mortgage rates between 20 and 50 basis points, with more recent estimates analyzing more current data generally falling toward the lower end of the range. A reduction in mortgage rates of around 20 to 50 basis points is, of course, considerably lower than the 2 percentage point rate change discussed earlier and thus should have a smaller effect on homeownership.

Given their methodology of examining a wide range of mortgages, analysts believe that the GSEs distribute their assistance broadly, which, in turn, helps to explain why the estimated mortgage rate reductions are relatively small per household. GSEs do not, for example, provide assistance solely to renters unable to become homeowners without GSE help.

Data and analysis on first-time homebuyers from the U.S. Department of Housing and Urban Development (HUD) are suggestive in this regard. The percentage of home-purchase loans financed by the GSEs that go to first-time buyers, particularly African Americans or Hispanics, is smaller than the percentage in the overall market and for Federal Housing Administration-insured (FHA-insured) loans (see Figure 2).

The GSEs' limited role in the first-time homebuyer market may reflect the fact that the majority of mortgages Fannie Mae and Freddie Mac finance have down payments exceeding 20 percent, even when borrowers have lower incomes. For example, HUD's analysis of the GSEs' funding of affordable loans, published in April 2002, found that "53 percent of the home loans purchased by Fannie Mae in underserved areas during 2000 had a more than 20 percent down payment, compared with [percent] of all home loans purchased by Fannie Mae."

Simulation evidence on cash assistance and down payment reductions

To provide additional context for the relationship between mortgage rate changes and homeownership, the simulations I reviewed compare the effect of mortgage rate reductions with other policy alternatives. Both underwriting and tenure-choice simulations review how reducing mortgage down payments cart affect the number of households that own homes. They find that down payment reductions have larger effects than mortgage rate reductions.

The underwriting simulations find that moving from a 5 percent down payment to a no-down-payment standard increases the percentage of all renters who can become owners by between 2 and 2.5 percentage points, and the percentage of African-American renters by between 1 and 2 percentage points.

Some of the tenure-choice simulations find even larger effects. In one case, moving from a percent to a zero percent down payment raised the probability of ownership by 4.5 percentage points for all households and percentage points for African-American households.

In this context, it is important to note that other aspects of the GSEs' operations beyond mortgage rate reductions can reduce the cost of buying a house. For example, the GSEs have special programs under which they will fund mortgages with down payments ranging from 3 to zero percent. (They have also relaxed other mortgage qualification standards.) Based on data provided in HUD's 2002 review of the GSEs' affordable loans, I calculate that in 1997, mortgages with down payments of equal to or less than percent equaled 2.5 percent of the home-purchase mortgages the GSEs financed. By 2000, the percentage had risen to 5.1 percent of the home-purchase mortgages the GSEs financed.

The earlier caveats about interpreting the simulation results apply for reviews of down payment changes as well. The absence of credit-risk data in these simulations, for example, may reduce the accuracy of the results. Those bearing the risk of the mortgage may want borrowers to have a higher credit score to compensate for the lower down payment.

The underwriting simulations also examine how providing lump-sum cash assistance to renters affects their ability to qualify for a mortgage. They find that such assistance can have a larger effect than either a down payment reduction or a mortgage rate reduction.

Cash payments starting around $5,000 have larger effects than other options on the ability of renting households to purchase a modestly priced home. One simulation found that a $5,000 payment increases the percentage of all renters who can buy a modestly priced home by 11 percentage points. (The percentage point increases are 13 and 7 for African-American and Hispanic households, respectively.) A payment of $10,000 per household has an effect almost twice as large.

It could be that the larger effect of the cash payment, as compared with the mortgage rate reduction, simply reflects a greater subsidy. To examine that potential, I assume that a direct cash assistance program would receive the same subsidy that Fannie Mae and Freddie Mac receive.

Analysts at the Congressional Budget Office estimate that Fannie and Freddie received $8.3 billion in average subsidy annually from 1995 to 2000 (the GSEs dispute this estimate). I assume the program would provide $10,000 per renting household. Thus, the program would serve 830,000 households a year. In three years, the direct subsidy program would assist 2.5 million renting households.

There were 105 million households in the United States as of 2000, according to the Census Bureau, with 69.8 million homeowners. A direct subsidy program serving 2.5 million households over three years would, all else equal, increase the homeownership rate by 2.4 percentage points. Of course, this estimate is not particularly sophisticated. Nonetheless, it suggests that a direct subsidy program to address households' inability to qualify for a mortgage could materially increase homeownership in a short period of time.

This direct-subsidy approach also appears consistent with President Bush's plan to increase minority homeownership. The Bush plan argues that the single biggest barrier to homeownership, particularly for minority households, is accumulating funds for a down payment. To help address that barrier, the Bush plan builds upon existing efforts to provide grants to households to pay for down payments and closing costs, among other efforts.

[FIGURE 1 OMITTED]

Figure 2

First-Time Homebuyers, 1997-1999

                                      As a Percentage of
First-Time        As a Percentage of  FHA-Insured Home-
Homebuyers        All Home Purchases    Purchase Loans

All                       41                  81
African American
 and Hispanic             11                  27

                  As a Percentage of
First-Time        GSE-Financed Home-
Homebuyers          Purchase Loans

All                       25
African American
 and Hispanic             3

SOURCE: AUTHOR'S CALCULATIONS BASED ON HUD DATA

Ron Feldman is assistant vice president with the Federal Reserve Bank of Minneapolis. The views expressed in this article are those of the author and not necessarily those of the Federal Reserve Bank of Minneapolis or the Federal Reserve System.