Home purchase lending in low-income neighborhoods and to low-income borrowers.

The long-standing debate about whether enough housing credit is being provided in lower-income areas has received renewed attention in recent discussions about how to implement the Community Reinvestment Act. In this article, we review the economic arguments underlying the debate and use 1993 data supplied under the Home Mortgage Disclosure Act (HMDA) to measure the extent of home purchase lending in lower-income and other neighborhoods as well as the extent of such lending to lower-income and other borrowers. We also summarize the 1993 HMDA data in the appendix to this article (see box "Highlights of the 1993 HMDA Data").

Highlights of the 1993 HMDA Data

Data reported by lenders under the Home Mortgage Disclosure Act cover applications and extensions of credit for home purchase loans, home improvement loans, and home refinancings. Here are some qualitative findings from the 1993 data.

* The number of reported loans and applications increased sharply from 1992 both because of a boom in refinancings and because of a rule change that raised the number of independent mortgage companies reporting.

* The use of government-backed mortgage programs for refinancings rose.

* Mortgage companies (including those affiliated with depository institutions) originated the majority of home purchase loans (and most government-backed home purchase loans) and the majority of home refinancings, whereas depository institutions originated most loans for home improvement and for multifamily dwellings (dwellings housing five or more families).

* Among would-be homebuyers grouped by racial or ethnic identity, black applicants were the most likely to seek government-backed mortgages.

* Among all home loan applicants grouped by racial or ethnic identity, black applicants were the least likely to file joint applications.

* Although most applications for home loans are approved, rates of denial for black and Hispanic applicants exceeded those for white and Asian applicants.

* The number of conventional home purchase loans extended to lower-income, black, and Hispanic households rose sharply from 1992.

* Among the home loans bought or securitized by government-sponsored enterprises, those securitized by Ginnie Mae were the most likely to involve lower-income households.

By most measures, the number of home purchase loans made by commercial banks and savings associations in lower-income neighborhoods is small compared with the number of such loans in higher-income neighborhoods. To some observers, this pattern of lending is readily understood within the context of the business of banking, where profit-seeking institutions strive to meet the demands of creditworthy borrowers. In this view (which we will call the "efficient markets" view), the low number of home purchase loans in lower-income neighborhoods reflects the relatively small number of credit-worthy borrowers and the relatively small supply of owner-occupied housing. To others, the disparity in such lending results from discrimination, neighborhood conditions, or both - circumstances that lead bankers, who may have little knowledge of lower-income neighborhoods, to erroneously conclude that these areas do not offer profitable lending opportunities. In these views - the first of which we will call the "discrimination" view and the second (regarding neighborhood conditions), the "externalities" view - the amount of home purchase lending in lower-income areas is too small.

The Community Reinvestment Act (CRA) requires the supervisory agencies of commercial banks and savings associations to encourage such lenders "to help meet the credit needs of the local communities in which they are chartered consistent with the safe and sound operation of such institutions."(1) For proponents of the efficient markets view, lenders already have ample incentives to seek all profitable lending opportunities and face few barriers to providing such credit everywhere; therefore such legislation should have little effect on lending because lenders are already doing what the law is encouraging them to do.(2) However, if the CRA forces lenders to make unprofitable loans, then the proponents of the efficient markets view would see the CRA as a burden on the banking system. For proponents of either the discrimination or the externalities view, lenders overlook safe and sound lending opportunities in lower-income or predominately minority neighborhoods; therefore, the CRA may have the power to change lending patterns to the benefit of both lenders and neighborhoods.

The debate over the appropriate amount of lending in low-income neighborhoods is perhaps best illustrated by the diverse views over whether to establish, as part of CRA monitoring and enforcement, a performance test for lenders that evaluates the quantity of their lending in the various neighborhoods of their local communities. Banks and savings associations frequently argue that such a test would result in nonmarket credit allocation, which is lending based on noneconomic criteria; such lending would more likely be unprofitable and would violate the original intent of the CRA. Community groups and others believe, however, that a loan performance test would simply encourage banks to extend a more economically appropriate amount of credit. These contrasting views of lending performance standards have persisted since the passage of the CRA. In an attempt to balance these concerns, CRA evaluations incorporate both subjective and objective measures of CRA compliance, and they take into account the characteristics and needs of an institution's community as well as the capacity of the institution and the constraints it faces.

A consideration of these issues in light of the 1993 HMDA data yields the following results, which will be discussed in detail in the balance of this article. The HMDA data for 1993, as for earlier years, show a relatively low number of home purchase loans in low-income areas. The data also suggest that although mortgage credit is extended in almost all neighborhoods, individual lenders vary greatly in the extent to which they are involved in low- and moderate-income neighborhoods. Many lenders make only a very small percentage of their home purchase loans in low-income neighborhoods or to low-income borrowers. Thus, a performance test based on the proportion of loans made either in low-income neighborhoods or to low-income borrowers could lead some lenders to change their lending behavior to satisfy the test, depending on the definition of their service area and on the particular characteristics of their market. A performance test that compares a lender's market share of home purchase loans in low- and moderate-income neighborhoods with the lender's share in higher-income neighborhoods would, however, yield a favorable outcome for many lenders who make loans in low-income neighborhoods. Such lenders generally have a larger market share in these neighborhoods because they face fewer competitors than is typical in their higher-income neighborhoods.

In addition, residents of low-income neighborhoods who obtain loans generally have incomes well above the median family income of their neighborhood, and low-income borrowers who purchased homes in 1993 frequently did not buy them in low-income neighborhoods. These findings illustrate the complications involved in targeting specific groups or neighborhoods for special efforts, and they suggest that (1) a lending test focusing exclusively on low-income neighborhoods may not necessarily encourage lending to low-income individuals and (2) a test focusing exclusively on low-income borrowers may not necessarily encourage lending in low-income neighborhoods. Furthermore, lending to people with low incomes may not be the same as lending to people with limited financial resources (for example, retirees may have low incomes but significant holdings of financial assets, and newly graduated professionals may have small current incomes but substantial future incomes).

However, lending in lower-income neighborhoods does more frequently result in loans to moderate-income borrowers than lending in other neighborhoods, and lending to lower-income borrowers does more frequently result in loans in lower-income neighborhoods than lending to borrowers with higher incomes. The current CRA proposal attempts to balance concerns about lending to lower-income neighborhoods with concerns about lending to lower-income borrowers, but it does not address the issue that some low-income borrowers may have substantial current or future financial resources.

A Brief History of the Community

Reinvestment Act

The Community Reinvestment Act of 1977 is intended to encourage commercial banks and savings associations to help meet the credit needs of the local communities in which they are chartered.(3) In adopting the CRA, the Congress reaffirmed the principle that depository institutions have an obligation under their charters to serve "the convenience and needs" of their communities by extending credit to all parts of those communities.

The CRA is directed primarily at four federal supervisory agencies-the Board of Governors of the Federal Reserve System, the Comptroller of the Currency, the Federal Deposit Insurance Corporation, and the Office of Thrift Supervision. The act calls upon the agencies to (1) use their supervisory authority to encourage each financial institution to help meet local credit needs in a manner consistent with safe and sound operation, (2) assess an institution's record of meeting the credit needs of its entire community, including low- and moderate-income neighborhoods, and (3) consider the institution's CRA performance when assessing an application for a charter, deposit insurance, branch or other deposit facility, office relocation, merger, or acquisition.

The debate on the bill that led to the CRA indicates that the Congress was concerned primarily with inner-city neighborhoods in general and with blighted and economically depressed areas in particular. The proponents of the CRA asserted that depository institutions were accepting deposits from households and businesses in these areas while lending mostly elsewhere and overlooking qualified loan applicants from the local community. They believed that the failure of depository institutions to take advantage of sound lending opportunities in these neighborhoods accelerated the process of economic decay and inhibited private revitalization efforts. Congressional supporters of the legislation viewed residential mortgage lending, including the provision of home improvement and rehabilitation credit, as particularly important for neighborhood stability and revitalization.

The legislative debate indicates that the Congress did not support nonmarket methods of credit allocation, such as quotas, to meet the credit needs of the local community. However, the Congress did not provide the regulatory agencies with any guidance in defining a low- or moderate-income neighborhood or a bank's "community," in assessing a community's credit needs, or in determining how well a particular institution is meeting those needs. To implement the CRA, the supervisory agencies adopted joint regulations that reflected two principles: The regulation should not require lenders to allocate credit; and financial institutions should be free to meet their CRA obligations in different ways to reflect the needs of their communities and their own capabilities.

To enforce the CRA, the regulatory agencies conduct CRA examinations of institutions and, as required by the statute, evaluate CRA performance during the application process for bank acquisitions, mergers, and other actions. The vagueness of the affirmative responsibility placed on lenders by the Congress has made determining compliance with the CRA difficult for the regulatory agencies. The agencies currently measure CRA compliance using twelve assessment factors, including activities undertaken by institutions to ascertain the credit needs of the community, to communicate the availability of credit services, and to provide credit to all geographic areas of their community (see box "Twelve CRA Performance Factors"). Most depositories receive a rating of satisfactory or better on their performance, and few institutions have had their applications for mergers or acquisitions denied. The CRA has, however, prompted institutions to undertake specific actions to enhance their CRA performance before or during the application process.

Twelve CRA Performance Factors

The federal supervisory agencies have traditionally considered the following factors in assessing an institution's record of performance under the Community Reinvestment Act:

* Activities conducted by the institution to ascertain the credit needs of its community, including the extent of the institution's efforts to communicate with members of its community regarding its credit services

* The extent of the institution's marketing and special credit-related programs in the community

* The extent of participation by the institution's board of directors in formulating the institution's policies and reviewing its performance with respect to the purposes of the CRA

* Any practices intended to discourage applications for types of credit set forth in the institution's CRA statement

* The geographic distribution of the institution's credit extensions, credit applications, and credit denials

* Evidence of prohibited discriminatory credit practices or of other illegal credit practices

* The institution's record of opening and closing offices and providing services at offices

* The institution's participation, including investment, in local community development and redevelopment projects or programs

* The institution's origination of residential mortgage loans, housing rehabilitation loans, home improvement loans, and small-business or small-farm loans within its community, or the purchase of such loans originated in the community

* The institution's participation in government-insured, -guaranteed, or -subsidized loan programs for housing, small businesses, or small farms

* The institution's ability to meet various community credit needs given its financial condition and size, legal impediments, local economic conditions, and other factors

* Other factors that, in the supervisory agency's judgment, reasonably bear upon the extent to which an institution is helping to meet the credit needs of its entire community.

After more than fifteen years of experience under the CRA, community organizations and depository institutions have expressed frustration with the process of enforcement. Community groups believe that the examination process fails to make meaningful distinctions between depository institutions that perform well and those that perform poorly. Depository institutions complain that CRA enforcement is too focused on process and paperwork and that the examination standards are unclear and inconsistently applied.

The Proposed CRA Regulation

Believing that the implementation of the CRA could be improved, President Clinton in July 1993 requested that the federal financial supervisory agencies reform the CRA examination and enforcement system. The President asked the agencies to refocus the CRA examination system on more objective, performance-based assessment standards with the goals of minimizing the burden of compliance, promoting consistency and even-handedness, and providing more effective sanctions against institutions with consistently poor performance.

In December 1993 the agencies published for public comment a proposal to revise the CRA regulation that would have substituted three performance-based measures - a lending test, an investment test, and a service test - for the twelve assessment factors currently used to measure CRA compliance.(4) In particular, the agencies proposed as the main part of the lending test a "market share" measure that would quantify an institution's CRA performance by comparing its market share of loans in lower-income neighborhoods to its market share in other neighborhoods.

The controversy generated by the December proposal was reminiscent of the arguments made during the consideration of the CRA in 1977. Many banks and savings associations believed that the market share test would result in forced credit allocation, while many community groups expressed concern that the formula did not take into account the complexity and effort involved in making different types of loans. In particular, community development loans, which can benefit groups with special credit needs but are time consuming and costly to originate, would have been treated in the same way as more standardized types of loans.

In response to the more than six thousand comments received on the December proposal, the regulatory agencies issued a revised CRA proposal in September 1994.(5) It, too, includes three performance-based measures of CRA performance, but the importance of the market share test and other quantitative standards is diminished (see box "The Proposed New Performance Tests"). The agencies stated that the examination process is inherently subjective and requires that performance be measured within the context of (1) a community's credit needs and (2) the capability of the lender. These two standards are referred to as the "assessment context." The benefit of the proposed new CRA regulation, according to the agencies, is that it would lower the costs of compliance and make CRA performance ratings more meaningful by spelling out, in greater detail, the process regulators would use to make a CRA evaluation.

The Proposed New Performance Tests

The September 1993 CRA regulatory proposal puts forth three tests by which to assess the CRA performance of depository institutions covered by the act: a lending test, an investment test, and a service test.

The lending test would employ assorted measures of lending activity for a variety of loan types, including home purchase loans. Among the assessment criteria are the geographic distribution of lending, the distribution of lending across different types of borrowers, the extent of community development lending, and the use of innovative or flexible lending practices to address the credit needs of low- or moderate-income individuals or areas.

The investment test would evaluate the extent of an institution's involvement with qualified investments. A qualified investment is an investment, deposit, or grant that benefits primarily low- or moderate income individuals, small businesses, or small farms or addresses affordable-housing needs.

The service test would evaluate the availability and responsiveness of an institution's system for delivering retail banking services and judge the extent of its community development services and their degree of innovation. Among the assessment criteria are the geographic distribution of an institution's branches and ATMs, the availability of alternative systems for delivering retail banking services in low- and moderate-income areas and to low- and moderate-income individuals, and the provision of community development services. A community development service is one that benefits primarily low- and moderate-income individuals or addresses the affordable-housing or other economic development needs of the community.

The CRA legislation places a heavy emphasis on the analysis of the geographic distribution of an institution's lending across its entire community. The CRA proposal would implement this legislative intent by classifying neighborhoods in a lender's service area as low-, moderate-, middle-, or upper-income.(6) A low-income area is defined as an area in which the median family income is less than 50 percent of the median family income for the broader area (such as a metropolitan statistical area, or MSA).(7) In a moderate-income area, the median family income is at least 50 percent and less than 80 percent of that for the broader area. In a middle-income area, the percentage ranges from at least 80 percent to less than 120 percent. And in an upper-income area, the percentage is at least 120 percent. As this article will show, these income divisions divide the population and the number of census tracts into groups of unequal size.

The September proposal also extends the evaluation of a bank's lending to encompass the distribution of loans across low-, moderate-, middle-, and upper-income borrowers, for which the income categories follow the same groupings as that for neighborhoods but rely on the borrower's income relative to that of his or her MSA. Thus, while continuing to place a heavy emphasis on the geographic distribution of an institution's lending, the agencies will favorably consider loans made to low- and moderate-income individuals.

CRA examiners consider a broad range of loan products, including all types of residential, consumer, and small business loans. This article focuses on home purchase lending, an important component of the proposed lending test. The HMDA data allow the empirical investigation of the nature and extent of this type of lending by the mortgage industry to different neighborhoods and different borrowers in all MSAs.(8)

As discussed below, the proposed regulation does not establish specific lending thresholds for obtaining a particular CRA rating; such lending is to be reviewed within an "assessment context." Moreover, a particular CRA rating would include the proposed investment and service tests-tests that are not considered in this article. Thus, the distribution of lenders measured by some aspect of their home purchase lending may not parallel the distribution of lenders by their CRA ratings.

A Review of the Business of Banking

Depository institutions, like other businesses, rely on the savings of households and businesses to develop and expand their activities. Because savers are seeking the highest risk-adjusted returns, depository institutions can attract additional funds and thus grow only when they offer a risk-adjusted return to savers that is higher than that from available alternatives. In turn, for lending activity to grow and prosper, expected revenues from lending must more than offset the expected costs of making loans.

One type of savings-capital (funds invested in a depository institution and not merely deposited)-is of special importance because the owners of capital-the shareholders in the institution-bear most of the risk of lending. If a borrower fails to pay back what was promised in the credit agreement, the shareholders usually must absorb the loss. The position of depositors differs from that of shareholders because depositors' savings are typically guaranteed by the government. If an institution fails, most depositors eventually get back their savings, whereas the shareholders lose some or all of their savings. Because their form of savings carries greater risk than that associated with deposits, shareholders expect their returns to exceed those of depositors.

The revenues from lending accrue primarily from the interest rates charged on the loans and any points or other fees assessed when the loans are originated. Although the lender knows at the time of loan origination how much revenue from points and fees it will receive on a particular loan, the timing and amount of revenue that will be generated by the stream of interest payments on the loan is much less certain: The borrower may repay the loan ahead of schedule (for example, a homeowner may refinance the mortgage or sell the home before the loan expires) or the borrower may fail to make the required payments and default on the loan. A lender, in determining the expected return from a loan, must assess the likelihood of each of these contingencies.

The cost of lending has at least four components. The first component is expenses required to pay staff and provide the physical facilities for operations. The second is expenses associated with processing loans, including origination activities (such as evaluating an applicant's credit history and ability to repay the debt as scheduled), the appraisal and survey of properties offered as collateral, and servicing activities (such as processing loan payments and monitoring borrowers who have fallen behind on their payments or are likely to fall behind). Although the amount of time and effort expended on origination, processing, and servicing activities vary considerably among loans depending on their type and complexity, all loans require a minimum level of such activities.

The third component is the expenses associated with possible loan default, often referred to as credit risk.(9) When a lender extends a loan, it knows that there is some probability that the borrower will not repay the loan or will significantly delay payment. In some cases of default, the lender is able to take action against the borrower-for example, by foreclosing on assets serving as collateral on the loan. If the collateral has sufficient market value, the lender may not suffer a loss on the loan. If the collateral is impaired or has declined in value, or if foreclosure expenses are too high, then the lender will incur a loss.

A lender that makes a large number of loans knows that a small percentage will not be repaid as scheduled. In estimating this risk, lenders often can rely on their past experience with similar types of credit. If they have had considerable experience with a particular type of lending, they can use statistical techniques to evaluate the credit risk posed by individuals with different characteristics and to establish reserves against possible losses. For unusual or large loans, they must rely on their judgment and experience.

The fourth component of lending costs is the expense associated with funding loans. The bulk of the funds used for loans and related activities comes from depositors; the remainder comes from shareholders who contribute capital. The interest rates paid on deposits and the returns on shareholders' capital must be high enough to attract and retain funds from both sources so that lending can take place.

How the CPA Might Affect

the Business of Banking

Economic theory offers at least three perspectives from which to consider how the CRA might affect the business of banking: (1) the lending market is efficient, (2) the lending market embodies illegal discrimination, and (3) the lending market is rendered socially inefficient by "externalities." The efficient markets view concludes that the CRA is a "tax" on the banking system, whereas the latter two views may, though not always, support the position that the CRA is a useful public policy tool that ultimately benefits lenders as well as borrowers and their neighborhoods.

The Efficient Markets View

The HMDA data indicate that home purchase loans are extended in almost all neighborhoods but that the loan volume in low- and moderate-income neighborhoods is smaller than in middle- and upper-income areas. Because financial institutions are present and able to lend in nearly all neighborhoods, an adherent of the efficient-markets view would conclude that the low volume of lending in low- and moderate-income neighborhoods simply reflects the lower returns of, and the lower demand for, such lending.

In this view, any regulation that encourages additional lending, including the CRA, is equivalent to a tax because it requires a depository to make a loan that is expected to yield, at best, an amount less than that required to cover all expenses, including the return expected by shareholders. As with all taxes imposed on a particular sector of the economy, the effect of higher taxes is to encourage economic activity to move to other, lower-taxed sectors of the economy. Thus, if the CRA is indeed such a tax and if all other things are held equal, then overall lending by CRA-covered depository institutions will decline (even though lending in lower-income neighborhoods may increase) relative to lending by institutions not covered by the CRA.

The Discrimination View

Economic theory suggests that systematic differences in treatment based on race or other inherent attributes of prospective borrowers (that is, illegal discrimination) may arise from two distinct sources: (1) prejudice or (2) information-based efforts to enhance profitability. The economic theory discussing prejudicial discrimination suggests that some individuals or firms are willing to forgo profitable exchanges to satisfy a "taste for discrimination" against a specific group.(10) The theory also suggests that illegal discrimination of this type will disappear over time as long as entry into markets is possible. According to this theory, firms that act on the basis of prejudice will eventually be driven out of business because other firms that do not discriminate will expand and take advantage of profitable opportunities left open by the firms that are discriminating. However, the length of time required for market participants to identify and take advantage of profitable opportunities is not specified by this theory.

If many lenders engage in prejudicial discrimination against minorities, perhaps reflecting a historical, societywide bias, then they may not extend credit in low-income neighborhoods because of the high proportions of minority residents in these neighborhoods.(11) In this case, the CRA is likely to increase lender profitability if it works with fair lending laws and forces lenders to lend to minorities who are good credit risks but who would have been denied loans because of illegal discrimination.

However, discrimination against minorities, against neighborhoods with high proportions of minority residents, or against low-income neighborhoods may be "information-based" discrimination.(12) Under this theory, group membership is perceived by the creditor as providing valuable information. That is, lenders may use an applicant's ethnic or racial status or neighborhood as a measure of the expected returns on a loan and thus require applicants with certain racial or ethnic identities or from certain neighborhoods to meet higher underwriting standards to qualify for credit. In this theory, market competition does not ensure the eventual elimination of information-based discrimination because such activities may be profitable, albeit illegal. Over time, however, the demand for loans from the groups subject to discrimination may decline, and such groups may develop and use alternative methods of financing home purchases (for example, the use of seller-financing or of government-backed lending programs). The loan demand may not return even if, at some point, lenders decide to seek the business of these groups.

As with prejudicial discrimination, the CRA combined with fair lending laws might be an effective, although indirect, mechanism for addressing information-based discrimination. But unlike the case of prejudicial discrimination, the CRA may lower profitability in the short-run if information-based discrimination can yield higher returns.

The Externalities View

An externality exists when the decisions and actions of one market participant affect the well-being of another without the decisionmaker's either bearing the full costs of, or reaping the full benefits from, those actions. For example, within a neighborhood, the failure of some homeowners to maintain their properties can have spillover effects on the prices of surrounding, well-maintained properties. The homeowners who allow their properties to decline suffer only some of the costs of their actions-their neighbors suffer some as well. And those who do maintain their properties likewise do not receive all the benefits of that choice-their neighbors receive some benefit as well.

A lender's assessment of the risks of, or returns to, mortgage lending in a neighborhood will be influenced by the condition of the properties there. If others are lending in the neighborhood at the same time, a lender may have greater confidence that the externalities will be positive because it knows that other lenders and borrowers are committed to maintaining or improving property values. Thus, a regulation that encourages more concentrated or coordinated lending in a particular neighborhood may raise the profitability of all lending in that neighborhood.

Many factors may account for the relatively lower volume of loans in lower-income neighborhoods, but a consequence of the lower volume is that lenders tend to have less experience, and thus higher costs, when evaluating the risks of lending in lower-income areas. If the information needed to assess the risks of some borrowers is costly to acquire, and the benefits of obtaining the information accrue only partly to the lender acquiring the information (and partly to competitors who do not seek costly information and therefore have, other things being equal, a lower cost structure), then lenders are less likely to invest sufficiently in acquiring the information; in other words, an information externality exists."

One information externality that may affect lending primarily in low-income areas arises from the paucity of real estate transactions in those areas.(14) If, in the case of single-family real estate, low- and moderate-income census tracts have fewer homes, fewer sales, and a more heterogenous housing stock than do upper-income areas, then establishing the value of homes for use as the collateral for mortgages will be more difficult. Lenders may hesitate to invest in learning fully about the market value of homes because keeping such information proprietary may be difficult.

Another information externality may be limited knowledge about governmental and other nonprofit programs for promoting lending in low-income neighborhoods. If subsidy programs can be used effectively to mitigate risk but acquiring information about such programs requires highly trained and highly paid personnel (who may be in short supply), then using these programs may be too costly for many lenders.

A lack of information about borrowers can also result in nonprice credit rationing and, as a consequence, some borrowers who are potentially able to repay loans may be unable to obtain credit. If a lender raises the interest rate to cover the cost of potential credit losses, it may attract a disproportionately large number of high-risk borrowers because low-risk borrowers, knowing that they have a relatively low probability of defaulting, are unwilling to pay higher interest rates. Consequently, the higher interest rates may fail to provide the overall returns expected by the lender, who may therefore find it more profitable only to make loans with lower interest rates to borrowers judged to pose a low risk. If a lender's knowledge about borrowers and perceptions of risk differ across neighborhoods, then neighborhood characteristics can be correlated with credit rationing.

If the CRA encourages the pooling of information about borrowers or neighborhoods, or the sharing of personnel, or joint lending activities, then, according to the externalities view, it may have a positive effect on profitability. Lending in low-income areas frequently involves the use of nontraditional lending criteria and subsidies and thereby makes risk assessment less standard and more complex. By sharing information as well as skilled and knowledgeable personnel, each lender may be able to lower the costs of using nonstandard lending criteria and to make profitable loans. Such cooperation might occur without the CRA because it is in the best interests of lenders as well as borrowers, but the CRA could make cooperation easier to the extent that the CRA sanctions and encourages such activities.(15)

If the CRA simply forces lenders to compete for less profitable loans, however, it may heighten the problems created by externalities. This result may arise if lenders believe they will not receive sufficient recognition for shared projects in their CRA evaluations and hence refuse to coordinate lending or share information. If the CRA encourages lenders to compete for "CRA loans" and undermines the sharing of information, then the profitability of lending may fall, and the CRA may again be thought of as a tax.

Characteristics of Census Tracts

Grouped by Neighborhood Income

As reviewed earlier in this article, CRA evaluations place emphasis on the distribution of a bank's services within its service area, with particular attention paid to low- and moderate-income neighborhoods.(16) The proposed CRA regulation would also have examiners evaluate a bank's performance within an "assessment context," which would include an analysis of information about the population, the housing units, and other characteristics of a bank's service area.

The basic unit of analysis for CRA evaluations is the neighborhood. One common way to define a neighborhood is as a census tract.(17) A few census tracts are primarily commercial or industrial in nature and have few or no residents. In our analysis of home purchase lending, we excluded census tracts if the 1990 census showed they had (1) no residents, (2) no owner-occupied housing, or (3) no reported median family income.(18)

As of 1990, low-income census tracts constituted only 7.5 percent of the roughly 42,700 census tracts in the United States (table 1).(19) Most census tracts fall into the middle- or upper-income categories.

Low-income census tracts are distinct from upper-income census tracts. First, median family income in the median low-income tract is only about 30 percent of the median income of the median upper-income census tract. Second, the population of the median low-income census tract is about two-thirds that of the typical upper-income tract; hence, the 7.5 percent of tracts that are low-income have only about 5.4 percent of the population of all tracts. Third, the median low-income tract has a proportion of minority residents that is more than eleven times greater than that found in the median upper-income census tract. Thus, an emphasis on lending in low- and moderate-income neighborhoods would likely result in a focus on credit to minority residents.

The demand for mortgage credit for owner-occupied homes in a census tract is influenced by the composition of the housing stock, the prevalence of families, and the level of population in the tract. Low-income tracts have few owner-occupied homes and only a small proportion of all the owner-occupied housing units in metropolitan areas because most residents in these tracts are renters. Only 2.2 percent of all owner-occupied housing units are in low-income census tracts, and the number of owner-occupied units in the median low-income tract is only one-fifth the number in the median upper-income tract. Moreover, the rate of family formation appears to be lower in low-income census tracts, judging by the ratio of families to households. (A household includes all people who occupy a housing unit whereas a family is a component of households and, as defined earlier, consists only of households with related persons.)

Other typical characteristics of low-income neighborhoods are high rates of poverty, housing vacancy, and unemployment.(20) In addition, the housing stock in such neighborhoods tends to be quite old, which may indicate that many properties in these neighborhoods are in poor physical condition. Before extending a loan, creditors often require that a property be in sound physical condition. Given these characteristics of low-income areas, potential borrowers there would seem less likely to seek home purchase loans and more likely to seek loans for home improvement and for rental properties.

In 1993, fewer than 2 percent of all census tracts had no reported home purchase loans. Census tracts without loans were excluded from further analysis because they have characteristics, such as fewer people and fewer owner-occupied housing units, that indicate they may be primarily commercial or industrial areas and because we cannot determine whether these census tracts are actually part of any particular bank's service area.(21)

1993 HOME PURCHASE LENDING

We now examinine the 1993 HMDA data on home purchase lending in terms of neighborhood, borrower, and lender characteristics.

Aggregate Lending by Neighborhood Income

The distribution of loans for the purchase of owner-occupied homes in 1993, as reported in the HMDA data, indicates that nearly 90 percent of the number of home purchase loans and about 92 percent of the dollar value of such loans are made in middle- and upper-income neighborhoods (table 2). For census tracts with at least one reported home purchase loan in 1993, the median low-income census tract had only five loans made within its boundaries compared with sixty-six in the median upper-income census tract. The number of loans extended, relative to the number of owner-occupied housing units, was 2.7 percent in the low-income tracts, compared with 7.2 percent in the upper-income tracts.

That the ratio of home-purchase loans to the number of owner-occupied housing units is much smaller in low-income neighborhoods than in upper-income areas may be viewed by some as evidence of discrimination in mortgage lending. However, this statistic by itself does not take into account important factors that may influence the volume of lending in these areas - property turnover rates, changes in home values, and the number of loan applicants who can qualify for credit.

Characteristics of Home Purchase Loan

Borrowers Grouped by Neighborhood Income

When evaluating a mortgage application, lenders often calculate two ratios of debt payment to income for the applicant: the "front-end" ratio, which includes only the debt payments associated with the home purchase (including property taxes and mortgage and home insurance), and the "back-end" ratio, which combines the proposed housing debt payments with the applicant's existing debt payments. Income, per se, is thus not necessarily a determinant of creditworthiness because a low-income borrower may be seeking a small loan for an inexpensive house whereas a high-income borrower may be seeking a large loan for an expensive house. Thus, a low-income borrower may have a lower payment burden than a higher-income borrower.

Income may, however, be a proxy for the ability of a household to save sufficient funds for a down payment on a home. Thus, there may be an income threshold below which relatively few borrowers can qualify for a mortgage regardless of debt payment ratios.(22)

In low-income areas, where the median family income of the census tract is, by definition, less than 50 percent of the MSA median family income, the median income of borrowers relative to the MSA median family income (referred to here as the borrower's relative income) was 76 percent in 1993 (table 3).(23) In moderate income tracts, where the relative income of families ranges between 50 and 80 percent of the MSA median, the relative income of borrowers was about 80 percent. Thus, for lower-income neighborhoods, a large drop in the median income of residents of a neighborhood seems to yield only a small drop in the relative income of borrowers, suggesting that financial institutions lend to borrowers who meet an income threshold regardless of neighborhood income.

The average borrower income is generally much higher than the median, even for low-income census tracts, because some borrowers have very high incomes. For borrowers in low-income census tracts, the average relative income was 113 percent (not shown in table), almost 50 percent higher than the median relative income of these borrowers. Overall, the median relative income of all home purchase loan borrowers was 109 percent compared with the average of 148 percent (not shown in table).

The loan-to-income ratio for the median borrower in each neighborhood income group is fairly uniform, ranging from about 189 percent to 200 percent; the narrow range perhaps reflects underwriting standards that do not vary significantly with neighborhood characteristics. Although the loan-to-income ratio is not generally calculated by lenders when underwriting a mortgage, it is directly related to the debt-to-income ratios described above.

A commonly used underwriting ratio is the loan-to-value ratio. HMDA data do not include the value of the home purchased by the mortgage borrower. However, the census data provide, for each census tract, the median value of owner-occupied homes in a census tract. Dividing the median-income borrower's loan amount by the median home value for the census tract of the property related to the loan provides a rough calculation of the loan-to-value ratio and indicates that loan-to-value ratios for the typical borrower are significantly higher in lower-income neighborhoods.(24) This rough approximation also indicates that-the-typical borrower in a low-income neighborhood made little or no down payment on the home purchase, while the typical borrower in an upper-income neighborhood made a down payment roughly equal to 20 percent of the home value.

The high loan-to-value ratio in lower-income neighborhoods may partly reflect the widespread use of government-backed mortgage programs in these neighborhoods (see appendix table A.4). New borrowers using these programs often have loan-to-value ratios close to 100 percent. A high loan-to-value ratio may also indicate a high proportion of first-time homebuyers. These buyers may have few financial assets available for a downpayment but relatively high incomes, and thus they tend to locate in neighborhoods with lower house prices.

Neighborhood Income and Borrower Income

Most borrowers in low-income neighborhoods do not have low incomes (table 4). In 1993, only 23 percent of the borrowers in low-income neighborhoods had low incomes, whereas more than 46 percent had middle or upper incomes. Perhaps surprisingly, the proportion of upper-income borrowers in the low-income census tracts was about equal to the proportion in the moderate-income census tracts (about 21 percent).

Similarly, most low-income borrowers (like borrowers in other income groups) purchased homes in middle- and upper-income neighborhoods (table 5). In 1993, 72 percent of low-income borrowers purchased homes in middle- or upper-income census tracts. Even middle- and upper-income neighborhoods have homes with modest prices, and many low-income borrowers may prefer to purchase homes in these neighborhoods because of the neighborhood amenities. In addition, some of these borrowers may be older, retired individuals with lower incomes but with substantial wealth, which allows them to purchase more expensive homes in middle- and upper-income neighborhoods.

The high proportion of middle- and high-income homebuyers in low- and moderate-income neighborhoods suggests that the additional emphasis on lower-income borrowers in the proposed CRA regulation may lead some CRA-covered lenders to change their lending strategies (for example, by participating in government-backed lending programs or by working more with nonprofit organizations). However, whether a loan to a high-income individual in a low-income neighborhood or a loan to a low-income individual in a upper-income neighborhood better fulfills the intent of the Congress when the CRA was enacted is unclear because, as mentioned above, the legislation is vague.

Distribution of Lenders by Neighborhood

Income and Borrower Income

Lending tests for evaluating CRA performance can be structured in many ways, but proposals for such tests often include the proportion of a bank's loans in low-income census tracts, the proportion of a bank's loans to low-income borrowers, or the bank's market share in lower-income neighborhoods relative to its share in higher-income neighborhoods. The current CRA proposal suggests that the examiner consider these ratios during a CRA performance examination but does not specify thresholds that would be associated with particular levels of performance.

The 1993 HMDA data provide an opportunity to examine the distribution of lending institutions across these ratios. The lending tests used in any regulation would be based on bank service areas. In using the HMDA data, our analysis assumes that a bank's service area corresponds to the census tracts where the bank originated home purchase loans in 1993. Furthermore, the HMDA data presented here represent a national distribution of lenders and thus may not reflect the distribution in any given market.

Distribution of Lending Activity

in Neighborhoods Grouped by Income

In 1993, only a small percentage of all home purchase loans (1.1 percent) were made in low-income neighborhoods (table 2). Not all lenders were equally active in low-income neighborhoods; in fact, most depository institutions covered by the CRA made few or no loans in low-income census tracts (table 6).(25) Given the relatively small number of low-income census tracts, many of these lenders may not have had such tracts within their bank service areas.

Most CRA-covered lenders were focused on middle- or upper-income neighborhoods, with about 88 percent of these lenders originating more than 20 percent of their home purchase loans in middle-income neighborhoods and about 62 percent originating more than 20 percent in upper-income neighborhoods. Few lenders originated a substantial portion of their loans in low-income neighborhoods. But the majority of CRA-covered lenders reported some involvement in moderate-income neighborhoods, with more than 55 percent of these lenders making at least 5 percent of their home purchase loans in these neighborhoods.

A large share of independent mortgage companies reported making less than 5 percent of their loans in low-income neighborhoods; like CRA-covered lenders, most mortgage companies were focused on middle- and upper-income neighborhoods. However, mortgage companies were more likely than CRA-covered lenders to originate more than 5 percent of their loans in moderate-income neighborhoods. As is discussed below, the wider use of the FHA loan programs by mortgage companies may allow them to make more loans to moderate-income borrowers in both low- and moderate-income neighborhoods.

Most credit unions, like other lenders, make few or no loans in low-income neighborhoods. Credit unions are more likely than other lenders to make no loans in other neighborhoods, particularly moderate-income neighborhoods. The relatively high proportion of credit unions with no loans within a neighborhood income group may reflect the fact that credit unions, unlike other lenders, can lend only to their members. Some credit unions may not have many members who seek to buy homes in certain neighborhoods.

CRA evaluations consider a lender's record in the context of its banking market because the needs of the market and the capability of a lender to meet those needs can sometimes be determined only by comparing an institution with its peers. One measure of the capability of a lender is its overall volume of loans for a particular product. Many lenders covered by HMDA do not specialize in home purchase lending and consequently report making few such loans.

In 1993, about half of the lenders who reported home purchase loans under HMDA made at least twenty-five such loans (they are referred to here as high-volume lenders). Using this benchmark as a crude definition of lenders that specialize in home purchase lending, we recalculated the distributions for the high-volume lenders. In this group, the proportion of lenders with no lending in low-income neighborhoods drops for all types of lenders (bottom half of table 6). The proportion for CRA-covered lenders, for example, drops from 74 percent to 57 percent. For moderate- and higher-income neighborhoods, the proportions of lenders reporting no loans generally decrease even more. Nonetheless, the typical high-volume lender's percentage of loans in low-income neighborhoods remained under 5 percent and in moderate-income neighborhoods remained within 5 percent to 20 percent.

A particular institution's position in the distribution of lenders, as measured by the proportion of loans in lower-income neighborhoods (or to lower-income borrowers) is sometimes suggested as one measure of CRA performance. But, as shown above, important in any such measure are the criteria that determine which lenders are included in a particular market. In addition, this type of measure could encourage unsafe and unsound lending unless appropriate consideration is given to the market definition, the credit needs of the neighborhood, and the lender's capacity to provide a particular type of lending.

Distribution of Lending Activity

among Borrowers Grouped by Income

The degree of lending among borrower income groups generally rises with the income of the group analyzed (table 7). Among CRA-covered lenders, the proportion of home purchase loans extended to low-income borrowers varies significantly. More than 12 percent of such lenders made more than 20 percent of their loans to low-income borrowers, while roughly 29 percent of these lenders reported granting no loans to such borrowers.

Mortgage companies, who generally sell the loans they make into the secondary mortgage market, may have more difficulty making loans to lower-income borrowers partly because they must meet secondary market underwriting guidelines. Such guidelines limit a lender's flexibility to extend credit to low-income borrowers, who more often have complex credit circumstances. Depository institutions often fund their loans directly and thus can vary underwriting standards to a greater extent.(26) The mortgage companies' lack of underwriting flexibility, relative to CRA-covered lenders, may partly account for the smaller share of mortgage companies extending more than 20 percent of their loans to low-income borrowers.

However, a larger share of mortgage companies reported extending more than 20 percent of their loans to moderate- and middle-income borrowers. Mortgage companies may be better able to reach moderate- and middle-income buyers because of their more frequent participation in government-backed lending programs. The programs impose uniform underwriting standards but provide greater latitude on the acceptable debt-to-income ratios and the amount of the borrower's funds needed to close a loan.(27)

Credit unions are less likely than CRA-covered lenders to make home purchase loans to low-income borrowers. More than three-fifths of the credit unions reported extending no loans to low-income borrowers. Unlike commercial banks or savings associations, credit unions can lend only to their members, and some credit unions may have few low-income members or few such members who seek home purchase loans.

A significant number of institutions make few or no home purchase loans to low-income borrowers. Proponents of the discrimination view and, perhaps, of the externalities view would likely argue that such lenders cannot be meeting the convenience and the needs of their communities because, as noted above, low-income borrowers can be found in almost all census tracts (table 5). Proponents of the efficient-markets view might respond that whether a particular lender provides a particular type of credit is less important than whether all qualified borrowers receive credit. In fact, in this view, if such lending requires specialized knowledge, not all lenders could profitably undertake such lending. Furthermore, although low-income borrowers are present in all census tracts, the number of such borrowers- might be few for some banks, particularly for those located in predominately upper-income areas and for those with many competitors.

As in the analysis of lender activity by neighborhood, we recalculated the distribution of lenders' share of lending by borrower income group after excluding low-volume lenders (bottom half of table 7). Here, dropping low-volume lenders produces sharp changes in the distributions. The proportion of lenders with no loans to low- or moderate-income borrowers falls significantly; and the proportion of lenders with more than 5 percent of their loans to low- or moderate-income borrowers rises.

[TABULAR DATA OMITTED]

Distribution of Lenders by Market Share

Although not given prominence in the current CRA proposal, one proposed standard of lending performance would compare a lender's share of reported loans (in various product lines) in low- and moderate-income areas of its delineated community to its share of reported loans in other parts of its service area. Precise evaluations of how individual lenders perform under such a standard are not possible because information is not available on the census tracts that constitute a lender's service area. As before, we can estimate market shares for each institution if we assume that the census tracts in which a lender reported home purchase loans under HMDA are the same as the census tracts the institution would identify as constituting its service area. This assumption is probably reasonably accurate for larger lenders but may be less so for smaller lenders, who may be more likely to extend home purchase loans in areas outside their service area (say, for instance, to long-standing customers who move to a new residential development in a suburb of a metropolitan area).

The 1993 HMDA data indicate that for a significant proportion of lenders, the market share of home purchase lending in low- and moderate-income census tracts exceeded their market share in middle- and upper-income neighborhoods (table 8). For example, 59 percent of the CRA-covered lenders had lower-income market shares that exceeded their higher-income shares (table 8, memo item). For many institutions, the lower-income neighborhood share exceeded the higher-income neighborhood share by a wide margin. The bulk of the institutions with low market share in low- and moderate income neighborhoods are low-volume lenders who made no loans in low- and moderate-income neighborhoods. When these institutions are excluded (bottom half of table 8), most lenders had higher market shares in lower-income neighborhoods than in higher-income neighborhoods.

[TABULAR DATA OMITTED]

The relatively high market shares in lower-income census tracts result from the smaller number of competitors in these neighborhoods (table 9). The number of competitors in upper-income neighborhoods is typically four to five times higher than the number in low-income neighborhoods.

9. Median and mean number of home purchase lenders in
each lender's "lending area," calculated aver all such
lenders reporting under HMDA, by census-tract income
group, 1993

Measure            Low   Moderate   Middle   Upper   All

Median..........    7       15        25      32      25
Mean ...........    8       16        26      34      27

  Note. For each home purchase lender, the census tracts in which the
lender made home purchase loans in 1993 were identified (the "lending
area") and grouped by income level. The total number of lenders that made
home purchase loans in those tracts in 1993 was then counted, and a median
and mean number of lenders in each lender's lending area, grouped by
income, was calculated. Finally, the median value of all lenders' median and
the mean value of all lenders' mean values were calculated and are reported
here. Excludes loans for which borrower income was not reported. See
general notes to table 1 and 2.
  Source. FFIEC.

SUMMARY

The volume of home purchase lending in low-income neighborhoods and to low-income borrowers in the United States is small and is generated by only a small proportion of lenders. The questions arise, is the small volume of such activity appropriate, and if not, how can it be raised?

Safe and sound lending practices require all lenders to evaluate the condition of properties offered as collateral for a prospective loan and to evaluate the creditworthiness of the prospective borrowers. A proponent of the efficient markets view of lending would see the characteristics of low-income neighborhoods as recorded in the census as partly explaining the low volume of home purchase lending and the relatively small number of lenders active in these neighborhoods. From this perspective, the small population, the high proportion of renters, and the often distressed nature of these neighborhoods, together with the requirements of safe and sound lending, translate into few opportunities to make profitable home purchase loans.

The efficient markets view suggests that the small amount of lending to minority households and the concentration of such households in lower-income neighborhoods are a result of their generally lower incomes and lower holdings of financial assets. A nondiscriminatory use of underwriting criteria that reflect the credit risk or the cost of extending credit to individual applicants will result in a volume of home purchase lending in low-income areas that is generally smaller than anywhere else.

Given the characteristics of lower-income neighborhoods, a proponent of the efficient markets view might also argue that lending in these neighborhoods is likely to be a highly specialized activity and thus most efficiently pursued by only a few lenders. As long as mortgage credit is made available in a nondiscriminatory and competitive manner, an efficient markets proponent would not be concerned about the number of lenders providing such credit. Furthermore, evidence that low-income borrowers are found in all neighborhoods and that borrowers from all income groups purchase properties in low-income neighborhoods suggests that a diversity of borrowers fall under the regulatory rubric of "low-income." An efficient markets proponent would argue that the market is best suited to determine which lenders and how many are needed to serve this diverse group of borrowers.

Finally, efficient markets imply that the CRA, if it alters bank lending, is a tax. Further, it may be a particularly onerous tax because it does not apply to all mortgage lenders and because it makes lending in low-income neighborhoods a responsibility of all commercial banks and savings associations that have such neighborhoods in their service area, regardless of their ability to extend such credit.

A different interpretation of the status of home purchase lending in low-income neighborhoods and to low-income borrowers comes from those who believe discrimination is the explanation for lenders' behavior. As noted, lending in low-income neighborhoods is frequently equivalent to lending in neighborhoods with a high proportion of minority residents, and lending to low-income borrowers often involves lending to minorities. In this view, the underwriting process used by lenders unfairly discriminates against minorities, and consequently a small proportion of loans are made in lower-income neighborhoods. Moreover, in this view, both the neighborhood characteristics found in the census data and the performance of lenders are, in part, outcomes of an unfair process.

From the discrimination view, the CRA may work with fair lending laws to push banks and savings associations to make loans that are profitable but overlooked because of the racial or ethnic composition of the neighborhoods. If discrimination is information-based - that is, the racial or ethnic characteristics of the neighborhood or of the borrower provide a low-cost method of screening loan applicants for potential returns - then the CRA may work with fair lending laws to overcome such illegal behavior, but it also may lower the profitability of commercial banks or saving associations.

A third interpretation of the data presented here is based on the view that certain market impediments, or externalities, are responsible for the small amount of lending in lower-income neighborhoods and to lower-income individuals. Proponents of this view may agree with the efficient markets view that the current level of home purchase lending in lower-income neighborhoods is consistent with characteristics of these neighborhoods, but they argue that the weakening or elimination of eternalities would allow the volume of profitable loans to rise.

From the externalities perspective, a smaller number of lenders in lower-income neighborhoods and a smaller number of loans to lower-income borrowers would be signs that information about the riskiness of this lending may be difficult to obtain and that such lending would be greater and more profitable if lenders shared information and coordinated lending activities. Moreover, lending to low-income borrowers - which often requires knowledge of government mortgage programs and other vehicles for qualifying lower-income households for mortgages - may also be a highly skilled endeavor and thus difficult for all CRA-covered lenders to pursue simultaneously. Thus, if the CRA encourages such coordination, then it may encourage additional profitable lending. If, however, the CRA merely encourages competition among lenders because of the need to make additional loans in low-income neighborhoods or to low-income individuals (that is, if loans resulting from coordinated activities are discounted in CRA evaluations), then a proponent of this view might also agree that the CRA is best considered a tax.

APPENDIX: SUMMARY

OF THE 1993 HMDA DATA

The Home Mortgage Disclosure Act of 1975 (HMDA) requires most depository institutions with offices in metropolitan areas to report the geographic location of the properties related to the home purchase, home refinancing, and home improvement loans that they originate or buy.(28) The data reported under HMDA are the only information on annual home lending activity at the neighborhood (census tract) level that is readily available to the public (see the box "Public Access to HMDA Data").

Public Access to HMDA Data

The HMDA data are made available to the public by the financial institutions covered by the act, by central data depositories (usually a public library, regional planning agency, or other public entity), and by the Federal Financial Institutions Examination Council (FFIEC). Under HMDA, each financial institution submits its data to its supervisory agency on a Loan/Application Register (HMDA-LAR) in a transaction-by-transaction format and makes an edited version of its LAR available to the public. The FFIEC processes the LAR data and prepares public disclosure statements for individual lenders; each lender in turn make the disclosure statements available in at least one office in each MSA in which it is located. The FFIEC also sends these disclosure statements and an aggregate report for each MSA to the central data depository in each MSA.

The FFIEC makes the HMDA data available for purchase in several media. Facsimiles of disclosure reports for the individual institutions and the aggregate reports for each MSA are available in paper form, in microfiche,and on CD-ROM. HMDA-LAR records and selected census data each tract in each MSA re available on PC diskettes, and such data for the entire nation are available on computer tape and CD-ROM. The sociodemographic data used to prepare the HMDA diclosure reports include data from the 1980 and the 1990 decennial Census of Population and Housing and annual estimates of the median four-person family income for each MSA from the Department of Housing and Urban Development; these data can also be obtained from the FFIEC.

The FFIEC also makes available a series of reports, drawn from he HMDA data, that were developed by the regulatory agencies to enhance their enforcement of fair-lending laws and assessment of lender efforts under the Community Reinvestment Act.

For information about the availability of data or to request data from the FFIEC, contact the HMDA Operations Unit, Mail Stop 502, Board of Governors of the Federal Reserve System, Washington, DC 20551. A copy of the HMDA data order form is available through of the HMDA Assistance Line, [phone].

Amendments to HMDA in 1989 and 1991 substantially increased the information that lenders must collect and the number of lenders required to report.(29) Under HMDA, covered lenders now report the disposition of home loan applications (for example, whether an application was approved, denied, or withdrawn) and on the race or national origin, sex, and annual income of applicants. They also report characteristics of borrowers whose loans are purchased. The type of purchaser of a loan must also be reported if the loan was sold the same year in which it was originated or acquired. As of 1993, a large number of additional independent mortgage companies became covered by the act; many of these firms are active lenders, often extending credit in dozens of metropolitan areas.(30)

This appendix summarizes the 1993 HMDA data. Beginning with the release of the 1994 HMDA data, these types of summary tables will appear each year in the Financial and Business Statistics section of the September issue of the Federal Reserve Bulletin.

The 1993 HMDA Data

For 1993, the 9,650 home lenders submitting HMDA data (table A.1) consisted of 5,300 commercial banks, 1,209 savings associations, 1,907 credit unions, and 1,235 mortgage companies, of which 962 were independent entities. The inclusion of additional independent mortgage companies in 1993 increased the number of such firms reporting data more than three-fold and accounted for most of the increase from 1992 in the overall number of reporters.

[TABULAR DATA OMITTED]

Using the 1993 HMDA data, the Federal Financial Institutions Examination Council (FFIEC) prepared 35,976 disclosure statements, each relating to a specific metropolitan statistical area (MSA), on behalf of the 9,650 covered institutions.(31) The FFIEC mailed these reports to lending institutions in July 1994 and to central data depositories in October. Both types of institution make the reports available to the public.

In 1993, lenders covered by HMDA acted on 13.6 million home loan applications and purchased 1.8 million loans. In comparison, in 1992 lenders reported information on 10.0 million home loan applications and 2.0 million purchased loans.

Of the 4.5 million applications for home purchase loans in 1993, 3.4 million, or 75 percent, were for conventional loans (table A.2). The remainder were for government-backed mortgages - loans insured or guaranteed by the Federal Housing Administration (FHA), the Department of Veterans Affairs (VA), or the Farmers Home Administration (FmHA).

[TABULAR DATA OMITTED]

Historically, relatively few homeowners have used either FHA-insured or VA-guaranteed loans when refinancing their existing mortgage. For instance, in 1992 less than 6 percent of the borrowers seeking to refinance applied for government-backed loans (data not shown in tables). The small share of refinancings using government-backed loans has reflected the relatively high cost and inconvenience of that type of transaction. Until recently, homeowners who refinanced their government-backed mortgages were usually those with equity sufficient to qualify for a conventional mortgage. In 1993, however, the use of government-backed loans for refinancing increased markedly, to 13 percent. This increase probably reflects the streamlined refinancing programs introduced by the government agencies, which allowed some homeowners with little equity to refinance at a low cost and with minimal paperwork.

In 1993, depository institutions originated about 48 percent of all home loans; independent mortgage companies or the mortgage company affiliates of depository institutions originated the rest (table A.3). Lending institutions tend to specialize in different types of home loans. In 1993, mortgage companies originated about 80 percent of the government-backed home purchase loans. Depository institutions, on the other hand, were the dominant source of home improvement loans and loans for multifamily dwellings.

[TABULAR DATA OMITTED]

Refinancing has been the most common type of home loan transaction in the past two reporting years, accounting for 55 percent of all reported home loans in 1992 and 60 percent in 1993. The extraordinarily high number of applications and credit extensions for refinancing in both years is attributable to low mortgage rates, the availability of no-fee loans, and more efficient processing of applications.(32)

Distribution of Applications by Applicant

and Census Tract Characteristics

In general, households with lower incomes are the most likely to use government-sponsored home loan programs, particularly those of the FHA and the FmHA. In 1993, 36 percent of home purchase loan applicants with incomes below the median family income for their MSA applied for government-backed loans; in contrast, only 15 percent of applicants with incomes of at least 120 percent of the median applied for such loans (table A.4, government-backed memo item).

[TABULAR DATA OMITTED]

The greater reliance of lower-income households on government-backed loans reflects several factors. Limits on the amount of FHA loan insurance make these loans unavailable to households seeking to buy more expensive properties; and the low down payment requirements and the ability to finance closing costs make them particularly attractive to lower-income households and to first-time homebuyers, who are likely to have fewer financial resources than others.

When examined according to the racial or ethnic identity of applicants, the data show that those who are black are more likely than others to seek government-backed home purchase loans. In 1993, about 44 percent of the black applicants for home purchase loans sought a government-backed mortgage; the comparable figure for Hispanics was 38 percent; for whites, 22 percent; and for Asians, 13 percent.

Finally, applicants for loans on homes in low-or moderate-income neighborhoods requested government-backed mortgage loans more often than those seeking homes in upper-income neighborhoods. Requests for government-backed loans also accounted for a higher share of all home purchase loan applications in neighborhoods with higher proportions of minority residents.

[TABULAR DATA OMITTED]

Most applications for home loans are filed by more than one person (table A.5). However, substantial differences in the frequency of loan requests by multiple applicants are found across racial groups. About 68 percent of requests by white or Hispanic applicants for conventional home purchase loans involved more than one applicant, whereas the corresponding proportion was about 44 percent for black applicants and 72 percent for Asian applicants.

Disposition of Home Loan Applications

In 1993, as in earlier years, most applications for home loans (regardless of purpose) were approved (table A.6). In 1993, lenders approved nearly

[TABULAR DATA OMITTED] 75 percent of the applications received for home purchase loans and about 82 percent of the applications for refinancings (and applicants either accepted or did not accept the loans). Applications that were neither denied nor approved were withdrawn or left incomplete by the applicant ("file closed").

High rates of approval for home loans generally, and for home purchase loans and refinancings in particular, are to be expected. Real estate agents and lenders often inform prospective loan applicants about the size of the loan for which they are likely to qualify. In addition, many home loan applicants have applied for mortgages in the past and are familiar with the basic underwriting guidelines used by lenders in making credit decisions.

Although most applications are approved, rates of denial vary among applicants grouped by their income and racial or ethnic characteristics and by the characteristics of the neighborhoods in which the properties involved are located. However, denial rates generally differ little among applicants grouped by sex, although applications filed by one woman are somewhat more likely to be approved than those filed by one man, and applications by men and women jointly generally are the least likely to be turned down (table A.5).

In 1993, 34.0 percent of black applicants, 25.1 percent of Hispanic applicants, 14.6 percent of Asian applicants, and 15.3 percent of white applicants were denied credit for conventional home purchase loans (table A.7). The comparable denial rates in 1992 were 35.9 percent for blacks, 27.3 percent for Hispanics, 15.3 percent for Asians, and 15.9 percent for whites. The lower denial rates in 1993 may stem from two sources: the lower mortgage interest rates prevailing in 1993, which, everything else equal, improved housing affordability; and the greater availability of special affordable housing programs (which often use more flexible underwriting standards) sponsored by both primary market lenders and secondary market institutions.

[TABULAR DATA OMITTED]

Differences in the distribution of applicants by income account for some of the differences in denial rates among racial or ethnic groups. Other factors are more important, however, because white applicants in each income category had a substantially lower rate of denial than the black or Hispanic applicants in that category (table A.8).

[TABULAR DATA OMITTED]

The pattern of denial rates found when applicants are grouped by income and by race or ethnic status is consistent with the increase in denial rates found for applicants from neighborhoods with greater proportions of lower-income or minority residents (table A.9).

[TABULAR DATA OMITTED]

Under HMDA, lenders also provide information on the reasons for denial. The most frequently cited reason for the denial of applications for conventional home purchase loans (regardless of the applicants' race or ethnic status) was a weak or nonexistent credit history; this factor was cited particularly often for blacks and relatively rarely for Asians (data not shown in tables). The most frequently cited reason for denial for Asians was an excessive debt-to-income ratio.

The extent to which racial discrimination may account for differences in denial rates among applicants from different racial groups is not known. The HMDA data provide little information about the creditworthiness of applicants and the characteristics of the properties that applicants seek to purchase or improve, and the data give no information about the specific underwriting standards used to evaluate each application. Thus, the information required by HMDA does not provide a solid basis on which to assess the fairness of the loan process.

Changes in Lending Volume

by Race and Income

In recent years, mortgage originators, as well as entities in the secondary mortgage market have initiated a wide variety of affordable home loan programs intended to benefit low- and moderate-income and minority households and neighborhoods. A comparison of the 1992 and 1991 HMDA data showed that these programs may have made a difference: Between those two years, the number of conventional home purchase loans extended to loan applicants with incomes below the median family income for their MSA went up 27 percent compared with only 10 percent for higher-income families.

A similar comparison of the 1992 and 1993 HMDA data indicates an even larger increase in the number of conventional home purchase loans granted to lower-income and minority applicants. The number of conventional home purchase loans extended to low-income households went up 38 percent from 1992 to 1993, while those extended to the highest-income households went up only 8 percent (table A.10, second column).(33) As for loans by racial or ethnic group, the number to black applicants went up 36 percent; to Hispanic applicants, 25 percent; and to white applicants, 18 percent. The changes in lending volumes were particularly large for lower-income black and Hispanic households. From 1992 to 1993, among low-income borrowers, the number of conventional home purchase loans went up 68 percent for black borrowers, and 50 percent for Hispanic borrowers, but only 36 percent for white borrowers (table A. 10, memo item).

A.10. Increase in number of conventional home purchase
loans from 1992 to 1993 for lenders reporting under
HMDA, by characteristics of borrower and census tract

Percent

                                                Excluding
Characteristics                  All lenders   newly reporting
                                                independent
                                               mortgage companies

All                                 24.6           16.5

           Borrower
Racial/ethnic identity
American Indian/Alaska native       18.7             7.3
Asian/Pacific                       15.0             6.5
Black                               43.9            35.8
Hispanic                            36.3            25.4
White                               24.6            17.5
Other                               74.4            64.1
Joint (white and minority)          25.8            17.8

Income
 (percentage of MSA median)
Less than 80                        46.2            38.4
80-99                               30.5            21.4
100-119                             25.5            16.2
120 or more                         16.0             8.2

Memo
Income less than 80 percent
of MSA median
American Indian/Alaskan native      30.8            22.1
Asian/Pacific Islander              38.7            28.6
Black                               75.2            67.7
Hispanic                            58.7            49.5
White                               44.0            36.4
Total                               46.2            38.4

       Census Tract
Racial/ethnic composition
(minorities as percentage
of population)
Less than 10                        26.0            18.2
10-19                               24.4            16.2
20-49                               21.0            12.6
50-79                               19.8            10.4
80-100                              17.1             7.9

Income
Low or moderate                     21.6            14.9
Middle                              23.8            16.1
Upper                               26.6            17.6

Note. Se text note 30 regarding 1993 expansion of HMDA coverage of
independent mortgage companies; see also notes to table A.4.
HMDA Data on Secondary Market Activity

HMDA Data on Secondary Market Activity

The large secondary market for mortgages - in which institutions buy and sell billions of dollars of mortgage loans or securities backed by mortgage loans each year - plays an important role in the US. housing market. The secondary market enables originators of mortgage loans to sell these otherwise relatively illiquid assets and obtain funds to extend new loans or to use in other ways.

Secondary market institutions generally do not originate loans but rather specify the underwriting guidelines that loans must meet to be eligible for purchase or securitization. Underwriting guidelines and related limitations on the size of loans that may be purchased vary among the secondary market purchasers.(34) As a consequence, for the loans that these institutions purchase or securitize, the characteristics of the borrowers and neighborhoods in which their properties are located can be expected to differ.

The 1989 amendments to HMDA require lenders to report the type of secondary market purchaser of home loans sold during that year (and the name, if it is a government-related entity). The HMDA data are currently the only publicly available source of information on the characteristics of borrowers whose loans are purchased by secondary market institutions and on the neighborhoods in which these borrowers reside.

Three government-sponsored enterprises (GSEs) dominate secondary market activity - the Federal National Mortgage Association (Fannie Mae), the Federal Home Loan Mortgage Corporation (Freddie Mac), and the Government National Mortgage Association (Ginnie Mae). Fannie Mae and Freddie Mac purchase or securitize mainly conventional mortgage loans, whereas Ginnie Mae only securitizes loans and only those that are government-backed. In 1993, the GSEs accounted for 72 percent of the roughly 7.4 million loans sold in the secondary market by lenders covered by HMDA (table A.11, top row).

[TABULAR DATA OMITTED]

Other institutions, including commercial banks, savings associations, insurance companies, and pension funds, are also active in the secondary mortgage market. Although the non-GSE institutions buy the same types of loans purchased by the GSEs, they provide an opportunity for lenders who originate nonconforming loans, such as jumbo loans, loans on mobile homes, and certain types of adjustable-rate mortgages, to sell their products as well.(35)

Because Ginnie Mae focuses on government-backed loans, it is more involved with home purchase loans extended to low- or moderate-income households than are other secondary market institutions.(36) For 1993, 55 percent of the loans securitized by Ginnie Mae were made to households whose incomes were less than the median family income of the MSAs in which they lived. The comparable figures were 30 percent for Fannie Mae and 29 percent for Freddie Mac. A similar pattern among the GSEs is found with respect to loans grouped by the income characteristics of the census tracts where the properties related to these loans are located.

Data Quality

The FFIEC member agencies have a program for the identification and resolution of errors in the HMDA information submitted by lenders. Over the past four years the quality of the HMDA data has improved. For the 1993 data, the proportion of loan records containing detected errors was less than 0.5 percent, down from about 4.4 percent in 1991.

To further improve the accuracy and completeness of the HMDA data, the Federal Reserve amended Regulation C in November 1994 to require lenders to update quarterly the HMDA information on their loan activity and to require most lenders to submit their data to the supervisory agencies in a machine-readable form (for example, on a computer diskette or magnetic tape). Under the new rules, which take effect January 1, 1996, lenders must record HMDA information about loan applications within thirty calendar days after the end of each calendar quarter in which a final action on an application is taken. This requirement will allow consumer compliance examiners to better monitor the quality of a lender's data.

Historically, the quality of HMDA data submitted in paper form has been far worse than that submitted in machine-readable form. Paper records are also more costly for the agencies to process. Before the November amendments, lenders that had 100 or more loan application records were expected (but not required) to submit data in machine-readable form rather than on paper; such lenders accounted for about three-fourths of all institutions. The new rule requires lenders that have more than 25 records to submit their data in machine-readable form and to edit their data before submission using free computer programs supplied by their supervisory agencies or comparable programs supplied by others.

Release for publication January 17

Industrial production rose 1.0 percent in December after a revised gain of 0.7 percent in November. The December increase was broadly based, with significant gains in mining and in durable and nondurable manufacturing. Unseasonably mild weather further depressed production at electric and gas utilities, however. Industrial production in December was at 121.4 percent of its 1987 average and 5.8 percent higher than it was in December 1993. Output grew at an annual rate of 5.4 percent in the fourth quarter, compared with 4.9 percent in the third quarter. The substantial growth in output in December boosted capacity utilization to 85.4 percent, its highest level since October 1979.

When analyzed by market group, the data show that the output of consumer goods rose 0.9 percent in December, after a similar increase in November. The production of durable consumer goods, propelled by another strong gain in automotive products, increased 1.9 percent; the production of other consumer durables, such as appliances, air conditioners, and carpets, also rose. The output of nondurable consumer goods, which had declined from June to October, advanced 0.8 percent in November and 0.7 percent in December, with gains in foods and tobacco, paper products, and drugs, soaps, and toiletries.

The production of business equipment, which advanced 1.1 percent in December, rose about 10 percent over the past twelve months. Industrial, information processing, and transit equipment all increased solidly in the past few months. The output of defense and space equipment, which had fallen substantially during the past few years, gained a bit in November and December.

The output of construction and business supplies changed little in December after strong gains in the preceding two months. The production of materials increased 1.3 percent. Once again, noticeable gains were spread widely among durable materials such as semiconductors, auto parts, plastics, and metals. Among nondurables, the output of chemicals and textiles advanced. The output of energy materials rebounded in December, despite continued weakness in the generation of electricity and the extraction of natural gas; the production of coal and crude oil rose.

When analyzed by industry group, the data show that manufacturing output rose 1.0 percent in December to a level 6.7 percent higher than that of a year earlier. Increases in production were pronounced among durables such as metals and machinery and equipment. Factories operated at 85.1 percent of capacity, just below the recent cyclical peak of capacity utilization in January 1989. The utilization rate in the primary-processing industries rose 0.7 percentage point, to 90.0 percent, which was above the cyclical highs reached in January 1989 and November 1978 but still below the 1973 average of 91.5 percent. Rates for primary metals, lumber, wood pulp, paperboard, plastics resins, petroleum refining, and rubber and plastics products remained relatively high. Utilization for advanced-processing industries rose 0.6 percentage point, to 83.1 percent, but remained 0.4 percentage point below its high in January 1989. Among advanced-processing industries, the operating rates for trucks, electrical machinery, and industrial machinery and computer equipment stayed high, but the rates for aircraft and space equipment, ships, instruments, and printing and publishing remained relatively low.

Because of continued mild weather, operating rates at utilities decreased further, to 84.5 percent in December compared with an average of 87.0 percent for 1994. Last June, when temperatures were unusually high, the utilization rate for utilities reached a high of 89.6 percent. Operating rates at mines rose 1.1 percentage points with a significant gain in output of coal.

Notice

An annual revision to industrial production, capacity, and capacity utilization was published on November 30, 1994. The revision to the production indexes affects data beginning in January 1991 and incorporates 1992 value-added proportions and revisions to monthly source data and seasonal factors. The revision to capacity and utilization incorporates the 1992 value-added weights along with available new data on physical capacity measures and investment. The capacity revision chiefly affects the individual series from 1991 forward; aggregate utilization may be changed slightly for earlier years to accommodate the introduction of 1992 weights.

Diskettes containing either historical data (through 1985) or more recent data (1986 to those most recently published in the G.17 release) are available from Publications Services, Mail Stop 127, Board of Governors of the Federal Reserve System, Washington, DC 20551 (202-452-3245). Files containing the revised data and the text and tables from this release are also available through the Economic Bulletin Board of the Department of Commerce; for information, call [phone].

A document with printed tables of the revised estimates of the series shown in the G.17 release is available upon written request to the Industrial Output Section, Mail Stop 82, Division of Research and Statistics, Board of Governors of the Federal Reserve System, Washington, DC 20551. (1.) Community Reinvestment Act of 1977, PL. 95-128, 91 Stat. 1147 (1977). (2.) The CRA may impose substantial compliance costs unrelated to lending, such as the costs of familiarizing staff with the requirements of the law and maintaining records to document for regulators actions taken to comply with the regulation. (3.) The act does not cover credit unions and other types of financial institutions. For a more expansive overview of the history of the CRA and of the issues associated with it, see Griffith L. Garwood and Dolores S. Smith, "The Community Reinvestment Act: Evolution and Current Issues," Federal Reserve Bulletin, vol. 79 (April 1993), pp. 251-67. (4.) "Community Reinvestment Act Regulations," 58 Fed. Reg. 67466 (1993) and 59 Fed. Reg. 5138 (1993). (5.) "Community Reinvestment Act Regulations," 59 Fed. Reg. 51232 (1994). (6.) Financial institutions may delineate their service areas using any method they choose provided that the definition does not reflect illegal discrimination, does not arbitrarily exclude low- and moderate-income areas (taking into account the institution's size and financial condition and the extent of its branching network), and consists of whole census tracts or block numbering areas.

In addition, for a retail bank, the service area is to include areas around its branches and deposit-taking ATMs in which it had originated or had outstanding during the previous calendar year a significant number and dollar amount of home mortgages and loans to small businesses, small farms, and consumers (if consumer loans are included by the bank in the CRA evaluation) plus any other areas equidistant from its branches and deposit-taking ATMs. (7.) According to the Bureau of the Census, a family consists of a householder plus at least one other person living in the same household related to the householder by birth, marriage, or adoption. A householder is, in most cases, the person or one of the persons in whose name the home is owned or rented.

An MSA typically consists of a central city with population of at least 50,000 plus the geographically larger jurisdiction, if any, in which the city is located and any other jurisdictions tied economically and socially to the city. (8.) HMDA applies only to lenders of a certain size or that receive a certain number of loan applications during the year and have a home office or branch in an MSA. Therefore the analysis presented here is limited to a subset of the mortgage industry. Nevertheless, the lenders covered by HMDA in 1993 accounted for about 73 percent of all home purchase loans made that year. (9.) Various factors affect the risks of and returns on lending by depository institutions across geographic areas. For additional details, see Board of Governors of the Federal Reserve System, Report to the Congress on Community Development Lending by Depository Institutions (Board of Governors, 1993). (10.) The first theoretical model of this type was developed in Gary S. Becker, The Economics of discrimination (University of Chicago Press, 1971). (11.) Prejudicial discrimination in lending may occur at different stages in the lending process: before the filing of an application (application screening), in the decision to approve or deny credit, in the setting of the loan terms, and in the level of service provided after a loan is extended. (12.) See, for example, Edmund S. Phelps, "The Statistical Theory of Racism and Sexism," American Economic Review, vol. 62 (September 1972), pp. 659-61. (13.) A detailed model of this phenomenon is presented in William C. Gruben, Jonathan A. Neuberger, and Ronald H. Schmidt, "Imperfect Information and the Community Reinvestment Act," Federal Reserve Bank of San Francisco, Economic Review, vol. 3 (Summer 1990), pp. 27-46. (14.) This argument is developed in detail in William W. Lang and Leonard I. Nakamura, "A Model of Redlining," Journal of Urban Economics, vol. 33 (1993), pp. 223-34. (15.) For examples of such cooperation, see Paul S. Calem, "The Delaware Valley Mortgage Plan: Extending the Research of Mortgage Lenders," Journal of Housing Research, vol. 4 (1993), pp. 337-58; and Richard G. Fritz and Rawi E. Abdelal, "Consortium Residential Lending and Community Reinvestment: An Analysis of the Atlanta Mortgage Consortium" (paper presented at the American Real Estate and Urban Economics Association, January 3, 1994).

A different method of overcoming these externalities would be to allow only a few lenders (or even one) to specialize in lending in targeted lower-income communities and have other lenders invest in these specialized lenders. However, to date the CRA has been interpreted as requiring all lenders to help meek-credit needs by lending directly in all parts of their local communities. (16.) A detailed description of characteristics of neighborhoods grouped by neighborhood income can be found in Glenn B. Canner, Wayne Passmore, and Dolores S. Smith, "Residential Lending to Low-income and Minority Families: Evidence from the 1992 HMDA Data," Federal Reserve Bulletin, vol. 80 (February 1994), pp. 79-108. In some cases, statistics in the 1994 article are not identical to those presented here because additional census tracts and MSAs were added in 1993, because the calculation techniques sometimes vary, and because loan product and census tract selection choices can differ. (17.) Census tracts are geographic areas drawn by the Bureau of the Census to be homogenous with respect to population characteristics, economic status, and living conditions. For geographic areas without census tracts, the proposed CRA regulation relies on block numbering areas, which the Bureau of the Census defines as a cluster of small geographic subdivisions signified by visible features such as streets, roads, railroad tracks and streams, and invisible features such as political boundaries. (18.) For 1993, 1,470 owner-occupied home purchase loans were reported in 282 census tracts that had a reported family median income but no owner-occupied housing units. The presence of owner-occupied home lending in these neighborhoods may reflect the conversion, since 1990, of commercial or industrial structures or public facilities such as schools to residential uses. In addition, some rental properties may have been converted to cooperatives or condominiums.

Also in 1993, 6,323 home purchase loans were reported in 914 census tracts that had no reported median family income. These tracts were recorded in the 1990 census with either no population or with a small number of people and, to protect the confidentiality of the respondents, the Bureau of the Census did not report a median family income. (19.) This count of census tracts includes only those in MSAs where a commercial bank, savings association, credit union, or mortgage company made at least one home purchase loan for an owner-occupied home in 1993 as reported under HMDA. The count excludes all census tracts in Puerto Rico. (20.) The poverty level for 1989 varied according to the number of persons in a household; the poverty level was $6,310 for a one-person household, $12,575 for a four-person household, and $23,973 for a nine-person household. (21.) Under the proposed regulation, lenders would have to identify the census tracts and block numbering areas in their service areas regardless of whether they made loans in these tracts. (22.) A similar point is made in Howard A. Savage and Peter J. Fronczek, "Who Can Afford to Buy a House in 1991?" Current Housing Reports (Bureau of the Census, July 1993), in which the authors estimate what percentage of current owners and renters could afford to buy a house. (23.) The borrower income is not the same as family income. A borrower may be a single individual or two or more unrelated individuals as well as a family. Moreover, income reported under HMDA may understate an applicant's actual income both because applicants sometimes do not reveal all sources of income to lenders and because lenders report under HMDA only that portion of the applicant's income relied on for the credit decision. (24.) The loan amounts are from the 1993 HMDA data, but the home values come from the 1990 census. As house prices increase, the loan size will also increase. To adjust for this mismatch in timing, we have inflated the 1990 home values by 9 percent.

There are three commonly used home price indexes. The Department of Commerce's new single-family, constant-quality unit index indicates that new home prices for the United States as a whole rose 5.8 percent from year-end 1990 to year-end 1993. The National Association of Realtors's existing single-family home price index shows an 11.8 percent increase over that period, while the Freddie Mac/Fannie Mae conventional mortgage home price index shows about a 9 percent increase. (25.) The CRA covers commercial banks and savings associations. Under the September CRA proposal, lenders have the option of including their affiliated mortgage companies in their CRA evaluation. Thus, for purposes of this analysis, we include mortgage companies affiliated with CRA institutions in the definition of CRA-covered lenders. (26.) For more detailed information about the influence of underwriting standards and the secondary market on community development lending, see Report to the Congress on Community Development Lending. (27.) A review of the FHA loan program can be found in Glenn B. Canner, Wayne Passmore, and Monisha Mittal, "Private Mortgage Insurance," Federal Reserve Bulletin, vol. 80 (October 1994), pp. 883-99. (28.) HMDA, which became effective in 1976, is implemented by Federal Reserve Board Regulation C (12 C.F.R. 203). For a more detailed description of the requirements of HMDA and a discussion of data from previous years, see Glenn B. Canner and Dolores S. Smith, "Home Mortgage Disclosure Act: Expanded Data on Residential Lending," Federal Reserve Bulletin, vol. 77 (November 1991), pp. 859-81; and Canner, Passmore, and Smith, "Residential Lending to Low-Income and Minority Families." (29.) The Financial Institutions Reform, Recovery, and Enforcement Act of 1989 contains the 1989 amendments to HMDA; the Federal Deposit Insurance Corporation Improvement Act of 1991 contains the 1991 amendments. The 1989 amendments to HMDA extended coverage for the first time to independent mortgage companies (mortgage companies not affiliated with a depository institution.) (30.) Before 1993, independent mortgage companies had to report only if they (or their parent) had at least $10 million in assets. Now, an independent mortgage company must also report if it receives 100 or more home loan applications during the year. (31.) The member agencies of the FFIEC are the Board of Governors of the Federal Reserve System, the Comptroller of the Currency, the Federal Deposit Insurance Corporation, the National Credit Union Administration, and the Office of Thrift Supervision. For HMDA-related matters, HUD also participates in FFIEC deliberations. (32.) With a no-fee loan, the borrower incurs no out-of-pocket expenses for either closing costs or discount points on the loan. Such loans are often written with a higher interest rate, and closing costs are typically added to the amount being financed. (33.) To provide the most appropriate year-over-year comparisons, the lending activity of the newly covered mortgage companies has been excluded from these calculations. (34.) Basic underwriting guidelines include the allowable debt-to-income and maximum loan-to-value ratios. Fannie Mae and Freddie Mac, as well as most other secondary market participants, establish their own guidelines for the conventional mortgage loans they purchase. In the case of Ginnie Mae, underwriting standards are established by HUD and by the VA. (35.) In this context, jumbo loans are conventional single-family mortgage loans that exceed the size limit on mortgages that may be purchased by Fannie Mae or Freddie Mac. In 1993, the limit was $203,150 for a single-family property; the limit was higher in certain locations, including Alaska, Guam, and Hawaii. (36.) Borrowers who use FHA-insured and VA-guaranteed mortgages differ from those who use conventional mortgages because HUD and the VA generally allow borrowers to qualify with more debt relative to income and to make smaller down payments than do conventional lenders. The more relaxed underwriting guidelines applicable to FHA and VA loans are often needed by families with more moderate incomes because, compared with families with higher incomes, they tend to carry relatively higher debt loads and have fewer assets available to make a down payment and pay closing costs.