Mortgage Lending in North Carolina After the Anti-Predatory Lending Law

Mortgage Lending in North Carolina After the Anti-Predatory Lending Law Final Report Cambridge, MA Lexington, MA Hadley, MA Bethesda, MD Chicago, IL September 14, 2004 Prepared by Kimberly Burnett Meryl Finkel Bulbul Kaul Abt Associates Inc. 55 Wheeler Street Cambridge, MA 02138

Acknowledgments This study was prepared under a contract between Abt Associates Inc. and the Mortgage Bankers Association. The authors gratefully acknowledge the contributions made by many people in completing this work. We wish particularly to thank Jay Brinkmann, Vice President of Research and Economics of the Mortgage Bankers Association, for his helpful comments on drafts of this report. Within Abt Associates we would like to thank Chris Herbert for his careful review of the draft report, Carissa Climaco for her work with census data, and Missy Robinson for her production of the report.

Table of Contents Executive Summary... ii 1. Introduction... 4 North Carolina s Anti-Predatory Lending Law...5 Preemption of State Laws by Federal Regulators...6 Previous Research...7 2. Purpose of the Study... 9 3. Data... 10 HMDA Data...10 Census Data...11 4. Control Group Selection... 12 5. Pre- and Post-Legislation Changes in Mortgage Activity... 15 Changes in Mortgage Activity By Market Segment and Loan Purpose...15 Changes in Mortgage Activity by Neighborhood Minority Composition...17 Changes in Mortgage Activity by Neighborhood Income Composition...21 6. Mortgage Originations by Federally Regulated Institutions... 23 Changes in Total, Prime, and Subprime Originations...23 Changes in Bank and Non-Bank Lending by Neighborhood Minority Composition...25 Changes in Bank and Non-Bank Lending by Neighborhood Income...26 7. Trends in Mortgage Application and Rejection Rates... 27 Changes in Applications and Denial Rates by Market Segment and Loan Purpose...27 Changes in Applications and Denial Rates by Neighborhood Minority Composition...30 Changes in Applications and Denial Rates by Neighborhood Income...33 Pre- and Post-Legislation Changes in Applications by Neighborhood Income Composition...33 8. Conclusion... 35 Bibliography... 40 Appendix... 41 Abt Associates Inc. Table of Contents i

Executive Summary Over the past several years, the problem of predatory lending has received increasing attention from public interest groups, lenders, and regulators. Until 1999, laws intended to protect borrowers from predatory lenders resided at the federal level. In that year, North Carolina became the first state to pass a law imposing more restrictive regulations on lenders for loans considered potentially predatory. Since then, many states and local governments have followed North Carolina s lead, passing similar anti-predatory lending laws. Critics of anti-predatory lending legislation assert that laws designed to curb predatory lending may also affect the availability of funds for legitimate subprime and, to a lesser extent, prime loans, cutting off access to credit for some borrowers. This may occur in part because a patchwork of various state laws makes it difficult and expensive for multi-state lenders to comply with regulations. This study examines changes in mortgage lending in North Carolina pre- and post-legislation at the neighborhood level to determine whether the law caused a type of redlining by reducing access to credit for borrowers in low-income and/or predominantly minority neighborhoods. We compare activity in North Carolina with lending trends in neighboring states with similar economic conditions that act as a control group. We examine changes in lending for all lenders, and then for prime and subprime lenders separately. For prime and subprime lenders combined, slower growth in mortgage lending was concentrated in low-income neighborhoods for both purchase and refinance loans and in predominantly minority areas, particularly for refinance loans. We find that growth in loan volumes among subprime lenders in North Carolina has not kept pace with growth in control group states, particularly among refinance loans. While the largest differences in growth rates between North Carolina and the control group states were in predominantly white neighborhoods and moderate-income neighborhoods, where there has been relatively slower growth in both purchase and refinance lending by subprime lenders, the only absolute declines in lending activity came in North Carolina s predominantly minority and lowincome neighborhoods. One exception to the relative slow growth in loan volumes among subprime lenders in North Carolina is in home improvement loans, where loan volumes grew faster in North Carolina than in the control states. The effects of the legislation were not uniform across types of subprime lenders. in loan originations by North Carolina s bank-owned and thrift-owned subprime lenders exceeded growth among these lenders in control states. Among non-bank subprime lenders, differences in growth between North Carolina and the control states were largest in predominantly white and moderateincome neighborhoods. Absolute declines were largest in predominantly minority and low-income neighborhoods. Declines in loan volumes in North Carolina relative to the control group appear to be attributable to relative declines in application volumes, not increases in denial rates. Indeed, North Carolina subprime lenders denial rates increased more slowly than those of subprime lenders in the control states. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law ii

Based on the comparison with neighboring states, it appears that the anti-predatory lending law has had an impact on loan originations in North Carolina relative to other states, and the impact has been felt in neighborhoods at all income levels and minority composition categories. However, the only absolute declines in lending came in low-income and predominantly minority neighborhoods. The study does not attempt to separate intended effects of the law to reduce predatory loans from unintended reductions in access to legitimate subprime loans. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law iii

1. Introduction Over the past several years, the problem of predatory lending has received increasing attention from public interest groups, lenders, and regulators. Although predatory lending is not well defined, several practices and loan features designed to strip owners equity from their homes are considered predatory under some circumstances. These practices include excessive loan origination fees, single premium credit life insurance, underwriting a loan without regard to the borrowers ability to repay, and loan flipping, in which a lender repeatedly refinances a loan with no benefit to the borrower. A key feature of predatory lending is that borrowers are charged higher fees and/or interest rates than necessary to account for the risk they represent. No studies exist, however, that establish the number of loans with predatory terms. Critics of anti-predatory lending legislation assert that laws designed to curb predatory lending may also affect the availability of funds for legitimate subprime and, to a lesser extent, prime loans. The term subprime generally refers to loans made to borrowers who have low credit ratings, high debtto-income ratios, or difficulties in documenting income history, but also includes very high loan-tovalue loans. The subprime mortgage market has increased access to credit for borrowers who do not qualify for conventional loans; however, subprime and predatory loans share the distinction of charging borrowers interest rates that are higher than conventional rates. To the extent that antipredatory lending laws also limit legitimate subprime lending, these laws will diminish the positive aspects of the development of the subprime market. In addition, numerous studies have demonstrated a concentration of subprime lending in high-minority and low-income neighborhoods; 1 given this concentration, anti-predatory lending laws might be expected to have a larger impact on lending volumes in these areas as well. The purpose of this study is to examine the impact of the anti-predatory lending law passed in North Carolina in 1999 on mortgage lending in that state. Specifically, this study examines changes in mortgage lending pre- and post-legislation at the neighborhood level to determine whether the law caused a type of redlining by reducing access to credit for borrowers in low-income and/or predominantly minority neighborhoods. We compare activity in North Carolina with lending in neighboring states that act as a control group and find that, among most types of loans, growth in loan volumes in North Carolina has not kept pace with control group states. We examine changes in lending for all lenders, and then for prime and subprime lenders separately. Some decline in loans those with predatory terms may be considered a desirable and intended outcome. Because our data do not contain any information on the loan s terms, we cannot separate intended effects of the law from unintended reductions in access to legitimate loans. Since there is no a priori estimate of the number of loans with predatory terms, it is not possible to conclude from this study that any of the reduction in lending was due to a decline in predatory loans, nor is it possible to conclude that any of the reduction in lending was due to a decline in legitimate loans. 1 See, for example, Bradford (2002), Herbert and Gruenstein (2000a), and Herbert and Gruenstein (2000b). Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 4

North Carolina s Anti-Predatory Lending Law Laws intended to protect borrowers from predatory lenders have, until recently, resided at the federal level. The Home Ownership Equity Protection Act (HOEPA), passed in 1994, imposes certain requirements on lenders for loans considered high-cost, such as additional consumer disclosures and restrictions on some loan features. More recently, many states and local governments have implemented laws that set a lower threshold for a high-cost loan and/or impose more restrictions on lenders for loans considered potentially predatory. North Carolina, in July 1999, was the first state to pass such a law. The law contains restrictions on loan terms for all home loans as well as a series of additional restrictions for loans considered highcost. High-cost loans are defined as residential loans under the conforming loan limit that meet one of the following three criteria: Fees exceed 5 percent of the loan amount for loans of $20,000 or more, or the lesser of 8 percent or $1,000 for loans under $20,000. Fees exclude: - Up to two bona fide discount points; - Prepayment penalties under 1 percent; - Escrows or fees paid to third parties. The annual percentage rate exceeds the yield on comparable Treasury securities by more than 10 percentage points, or A prepayment penalty more than 30 months after the loan closing or that exceeds 2 percent of the amount prepaid. The law prohibits the following for all loans: Financing up-front, single-premium insurance; Loan flipping, or a refinance with no reasonable tangible net benefit to the borrower; and Prepayment penalties for loans under $150,000. For loans defined as high-cost, the anti-predatory lending law adds the following: Balloon payments (a scheduled payment of more than twice the regular payment) are prohibited; Negative amortization is prohibited; Lending without proper consideration for the borrower s ability to repay the loan is prohibited; Home-ownership counseling is required; and Financing of fees or charges is prohibited (including points or fees to third parties). Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 5

Open-end credit plans, including home equity loans, are excluded from the definition of a home loan and not subject to these restrictions. The ban on prepayment penalties for loans under $150,000 and the no flipping rule became effective in October 1999; the remainder of the la w was implemented in July 2000. Since then, over 15 other states have also passed laws designed to prevent predatory lending, many of them using the North Carolina law as a model. Critics of state -level anti-predatory lending legislation argue that a patchwork of various state laws makes it difficult and expensive for multi-state lenders to comply with regulations. As a result, some lenders will withdraw from these areas to avoid having to tailor their lending to these state-specific laws. In addition, restricting loan terms will reduce consumer options for borrowing that will restrict their access to mortgage financing. Preemption of State Laws by Federal Regulators Several federal regulators have responded to the concerns that state laws prevent financial institutions from operating under consistent national standards. For example, the Office of the Comptroller of the Currency (OCC) issued a final rule, 2 effective February 12, 2004, that identifies the types of state banking laws that are preempted by federal regulation of national banks. 3,4 State anti-predatory lending laws are preempted, and therefore national banks are subject to federal laws, not state laws. In the case of North Carolina, federal laws are substantially less restrictive than state laws. The OCC articulated that its responsibility is to ensure that national banks can operate to the full extent authorized by federal law. The rule stressed the significance of consistent national standards, and expressed concern about the costs and interference that diverse and potentially conflicting state and local laws have on the ability of national banks to operate under the powers of their Federal charter. 5 Significantly, federal law also preempts state law for operating subsidiaries of national banks. In response to concerns that federal preemption would strip consumers of protection from predatory lenders, the OCC also adopted anti-predatory lending standards that govern national banks lending activities in the final rule. These standards prohibit national banks from making a consumer loan (loans for personal, family, or household purposes, including mortgages for owner-occupied properties) based primarily on the value of the collateral, and not on the borrower s ability to repay the loan. In addition, the OCC affirmed that national banks are prohibited from engaging in unfair or deceptive practices under the Federal Trade Commission (FTC) Act. The final rule notes that several practices considered predatory such as equity stripping and loan flipping can generally be considered to violate the FTC Act. The Office of Thrift Supervision (OTS) has not issued a similar document preempting state laws for federally regulated thrifts, but it has specifically stated that its federal regulation preempts important provisions of anti-predatory lending laws in Georgia, New York, and New Jersey. 2 3 4 5 Federal Register, Vol. 69, No. 8, January 13, 2004, pp. 1904-1917. State laws that are not preempted by federal regulation include those that relate to contracts, collection of debts, acquisition and transfer of property, taxation, zoning, crimes, torts, and homestead rights. National banks originated 29 percent of loans in North Carolina in 2002. Federal Register, Vol. 69, No. 8, January 13, 2004, p. 1908. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 6

Previous Research Several studies on the impact of North Carolina s anti-predatory lending law have been completed, in part because it is the state with the longest experience. Several different approaches have been taken, using various sources of data; however, all the studies have found that levels of subprime lending in North Carolina have declined relative to other states. One study concludes that this is a desirable outcome of the legislation; others use the same finding to point to a restriction in the flow of capital to higher-risk borrowers, most of whom would not have been victims of predatory lending. Harvey and Nigro (2002) use quarterly HMDA data from 1998-2000 to analyze levels of mortgage lending activity in North Carolina in comparison with neighboring states that have not implemented anti-predatory lending laws, including Georgia, South Carolina, Tennessee, and Virginia. They find that the number of subprime, prime, low-income, and minority originations all declined during the six quarters following enactment of the initial provisions of the predatory lending legislation compared with the previous six quarters. Further, the declines in originations were larger than those experienced by a set of four control group states (where subprime originations actually increased). They conclude that the law reduced the overall level of subprime mortgage lending activity, although the impact was not uniform. Minority applicants were disproportionately affected, as were lowincome households, although to a smaller degree. In addition, non-bank subprime lending levels dropped more than levels of subprime lending activity by banks. Using the same methodology but a longer period (average lending activity over and 2000-2002), 6 we obtained similar results: the growth in volume of prime, subprime, low-income, and minority originations in North Carolina all lagged growth in the same set of control states (Exhibit 1). The longer time period allows for complete adjustments to the predatory lending law. For example, some lenders may have stopped making certain types of loans in North Carolina because of uncertainty about the interpretation and implementation of the law, but returned after seeing the law in practice. s in mortgage volume growth rates between the control group and North Carolina were all sizeable, suggesting that effects of the predatory lending law found by Harvey and Nigro found were sustained through 2002. In addition, the differences were larger than those Harvey and Nigro found, probably due in part to North Carolina s deteriorating economic conditions in 2001-2002 (see Section 4). The largest difference was in subprime lending, where growth in average lending in North Carolina from to was 34.2 percentage points lower than in the control group. 6 Following Harvey and Nigro, we excluded loans with missing income information from the analysis entirely, and we excluded loans with missing race information from the minority lending analysis. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 7

Exhibit 1 Replication of Harvey and Nigro Analysis: Loan Originations by Market Segment Pre- and Post-Legislation Total Originations Subprime Prime Low Income Minority North Carolina avg. 210,951 23,904 187,047 23,838 25,764 avg. 234,349 24,398 209,951 14,622 26,419 Change 23,398 494 22,904-9,216 655 Control Group avg. 656,241 59,334 596,908 64,498 87,854 avg. 835,212 80,857 754,355 45,614 112,106 Change 178,971 21,524 157,447-18,884 24,252 Rates North Carolina 11.1% 2.1% 12.2% -38.7% 2.5% Control Group 27.3% 36.3% 26.4% -29.3% 27.6% -16.2% -34.2% -14.1% -9.4% -25.1% Harvey and Nigro -4.7% -17.2% -3.8% -3.5% -9.2% Note: Control group states include Georgia, South Carolina, Tennessee, and Virginia. Quercia, Stegman, and Davis (2003) use a proprietary database of securitized subprime loans from 1998 through 2002. The database includes information on loan terms as well as borrower credit scores, allowing an assessment of the change in access to credit for higher-risk borrowers. Like Harvey and Nigro (2002), Quercia, et al., also find that overall subprime originations declined more in North Carolina than in other states, and that this decline was entirely attributable to a drop in subprime refinance originations. Changes in purchase originations in North Carolina were similar to those in other Southern states. Quercia, et al., point out that the law is intended to reduce predatory lending, and that a decline in certain types of originations is to be expected. They examine changes in the number of subprime refinance originations with characteristics considered predatory: prepayment penalty terms of three years or longer (prohibited under North Carolina s law for high-cost loans and all loans under $150,000); balloon payments (prohibited for high-cost loans); and loan-to-value ratios (LTV) of over 110 percent. They use an LTV of over 110 percent to define loans that have no net tangible benefit to the borrower (prohibited in the no-flipping provision of the law). Quercia, et al., find a large decline in subprime refinance originations with these terms relative to loans with these terms in other states. They also find that changes in the number of subprime loans to high-risk borrowers (those with credit scores below 580) were similar to those of other states, but that subprime loans to borrowers with credit scores above 660 dropped significantly compared to trends in other states. They suggest that this may indicate that fewer low-risk borrowers are being steered into subprime loans. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 8

Last, Quercia, et al., dispute the idea that the anti-predatory lending law may have reduced the flow of capital for subprime mortgages in North Carolina, noting that interest rates in North Carolina have not changed relative to interest rates in other states. If the law reduced the flow of capital, an increase in the cost of capital would be expected. Elliehausen and Staten (2002) use data from nine members of the American Financial Services Association to compare subprime lending originations in North Carolina with those in a set of three control group states (Virginia, South Carolina, and Tennessee). The dataset they use includes loans originated between July 1995 and June 2000, and contains information on the loan terms, the characteristics of the borrower, and payment performance. They find a significant decline in subprime mortgage originations, particularly to lower-income borrowers. The effect on lower-income borrowers found is larger than that found by Harvey and Nigro (2002), who use a longer time period following full enactment of the law for their analysis (all of 2000 compared to only June of 2000 for Elliehausen and Staten). The difference in these findings may be the result of a large temporary dip in loans to lower-income borrowers that subsequently recovered to some degree. The study does not include any data at all after the date the law was fully implemented in July 2000. Their findings are therefore based on a few quarters during which mortgage activity may have been anomalous. Ernst, Farris, and Stein (2002) use 1999 and 2000 HMDA data to compare subprime originations for North Carolina with the rest of the nation. They find an overall decline in subprime lending relative to the rest of the nation, but that North Carolina was still among the most active subprime origination markets in the country. They estimate that the law saved borrowers more than $100 million through a substitution of loans with predatory terms for loans without such terms. Unlike Quercia, et al., their conclusion that the change in subprime lending was a positive development is based on broad assumptions about the proportion of subprime loans with predatory terms, and not on loan-level data that include loan terms. In addition, their estimate of savings to borrowers does not attempt to include the cost to consumers who may not have been able to get access to credit after the implementation of the law. 2. Purpose of the Study The purpose of this study is to examine the impact of North Carolina s anti-predatory lending law from a different perspective than that taken in earlier studies. This study examines changes pre- and post-legislation in North Carolina and nearby states at the neighborhood level. Specifically, we address the question of whether the law may have unintentionally resulted in a form of redlining by restricting the flow of credit to certain types of neighborhoods, such as those where a large share of the population is low-income and/or minority. As noted above, earlier studies have focused on overall levels of lending at the state level (Ernst, et al., 2002 and Quercia, et al., 2003), the county level (Elliehausen and Staten, 2002), or at the borrower level (Harvey and Nigro, 2002, and Quercia, et al., 2003). Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 9

Trends in originations, applications, and denial rates are compared for North Carolina and two control states: Tennessee and South Carolina. Harvey and Nigro (2002) used Virginia and Georgia 7 in addition to these two states; Elliehausen and Staten (2002) use Virginia, South Carolina, and Tennessee as their comparison group. North Carolina is compared with the control group by examining annual averages during a pre-legislation period () with annual averages postlegislation (). It is important to note that the data we use for this study does not include information on loan terms, such as interest rates, points and fees, LTV ratio, or prepayment penalty terms. Therefore, although we find that mortgage originations have declined in certain types of neighborhoods in North Carolina relative to the control group, we cannot conclude whether this decline was due to a drop in good loans versus a drop in bad loans (those with predatory terms). The next section of this report describes the data used in the analysis; Section 4 compares economic conditions in North Carolina and the control group states; and Section 5 reports findings on pre- and post-legislation changes in mortgage activity by market segment, loan purpose, neighborhood minority composition, and neighborhood income composition. Section 6 compares trends in mortgage activity by federally regulated institutions (banks) with those of state-regulated financial institutions (non-banks). Last, Section 7 examines the reasons for declines in lending volumes (relative to the control group) in North Carolina by analyzing trends in mortgage application and rejection rates. 3. Data In conducting this analysis, we use three data sources. These are data collected under the Home Mortgage Disclosure Act (HMDA); the subprime lender list developed by the Department of Housing and Urban Development (HUD); and 2000 decennial census data from the U.S. Census Bureau. HMDA Data HMDA data are used to analyze the pattern of mortgage originations and applications in North Carolina and in control group states. Under HMDA most mortgage lenders are required to report information on their home mortgage applications including the race, census tract, and income of borrowers and the action taken on the application. 8 Because HMDA reporting is required only for lenders that originate loans in metropolitan areas, our analysis is restricted to North Carolina s metropolitan areas. It is important to note that lenders are not required to report information under HMDA on second mortgages and home equity loans unless the lender classifies them as being used for refinance or 7 8 In Georgia, a predatory lending law took effect on October 1, 2002. Although the law was quickly preempted for thrifts by the OTS (in January 2003) and for national banks by the OCC (in July 2003), the law may have affected mortgage activity in that state. HMDA reporting is required only for lenders that meet certain annual loan volume and asset size requirements. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 10

home improvement. The limitation is an issue for analysis of subprime loans as subprime loans are often home equity loans. 9 If a home equity loan is used to refinance an existing home mortgage it should be reported under HMDA. However, if it is a second mortgage or if there is no existing primary mortgage that is being refinanced, lenders do not have to report these loans. As a result, some subprime home equity loans do not appear in the HMDA data. Despite this shortcoming, HMDA data are the best available on lending activity at a local level. HMDA data do not include direct information for identifying subprime loans, such as the interest rate. Therefore, subprime loans must be identified using the characteristics of the lender, not the loan itself. This is done using a list of subprime lenders developed by HUD. For each year since 1994, HUD has analyzed industry publications, lending data, and conducted interviews to determine which lenders originate primarily subprime loans. This list varies by year because of exits and entrances into the subprime lending market and changes in business strategy and market focus. The distinction between subprime and conventional lenders is not always obvious many lenders engage in both types of activity. If at least half of a lender s originations are subprime, the lender is included on HUD s subprime lender list. Given that some conventional lenders make subprime loans and some subprime lenders make conventional loans, the characterization of loans into a subprime category based on the lender originating the loan is only an approximation. It is important to note that over the last few years the distinction between prime and subprime lenders has become increasingly blurred. Primarily prime lenders have become increasingly involved in subprime lending, often by acquiring an existing subprime lending operation. This reduced ability to identify subprime loans in HMDA may result in an underestimate of subprime loan volumes in recent years. Importantly, since this trend should affect all market areas relatively equally, our comparison of trends across markets provides an implicit control for this problem with identifying subprime lending. Census Data A prime consideration of the study is the effect of the anti-predatory lending law on neighborhoods of different income and racial/ethnic composition. The neighborhoods used in this study are census tracts, which are areas defined by the Census Bureau to capture neighborhoods of about 4,000 in population. We use 2000 decennial census data to categorize neighborhoods into three groups based on their median income relative to the metropolitan area in which they are located: Low-income neighborhoods are those with median incomes below 80 percent of the area median income; Moderate-income neighborhoods are those with median incomes between 80 and 100 percent of the area median; and High-income neighborhoods are those with incomes above 100 percent of the area median. 9 Gramlich, Edward M., Tackling Predatory Lending: Regulation and Education. Speech given at Cleveland State University, Cleveland, Ohio, March 23, 2001. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 11

We also use 2000 decennial census data to assign neighborhoods to one of four categories based on their racial/ethnic composition: Predominantly white neighborhoods are defined as those where 10 percent or less of the population is minority; Mixed-majority neighborhoods are those where 11-25 percent of the population is minority; Mixed-minority neighborhoods are those where 26-50 percent of the population is minority; and Predominantly minority neighborhoods are those where more than 50 percent of the population is minority. The share of the population in North Carolina that is Hispanic is relatively small (about 5 percent), so we do not analyze the effects of the anti-predatory lending law on Hispanic neighborhoods separately from the effects on either predominantly minority neighborhoods or predominantly black neighborhoods. The HMDA data through 2002 report lending activity using 1990 census tract definitions, not the newly defined tracts used in reporting the 2000 census data. Therefore, we bridge the difference between the 1990-based census tract definitions used in HMDA with 2000-based tract definitions that provide the most up-to-date tract income and minority composition characteristics using files rele ased by the Census Bureau that provide information on the relationship between 1990 and 2000 census tract definitions. 10 Of North Carolina s 1990 census tracts, 71.5 percent are related to only one 2000 census tract. For the remaining 1990 census tracts with more than one possible 2000 census tract match, mapping software (MapInfo) is used to measure the distance between the centroid of a particular 2000 census tract and the centroid of 1990 census tracts. The 2000 census tract is assigned to the closest 1990 census tract. This procedure provides a good approximation of the 2000 census tract in 1990, although not an exact match. Census tracts in 2000 for the control group states are linked to 1990 tracts using the same process. 4. Control Group Selection Although North Carolina and the four-state control group used by Harvey and Nigro (2002) have many geographic, economic, and demographic similarities, there are important differences that may have affected growth in mortgage originations. The primary driver of mortgage originations, and particularly refinance originations, is interest rates (Megbolugbe and Cho, 1993). Interest rates across North Carolina, Georgia, South Carolina, Tennessee, and Virginia were similar during the period 1997-2002, differing by at most 23 basis points, in 2001 (see Exhibit 2). Note that some of the differences in interest rates are a result of variations in the size of the jumbo and subprime markets in each state, both of which have higher interest rates than other conventional single-family mortgages. For example, Virginia s consistently higher than average interest rates are likely a result of the large 10 See http://www.census.gov/geo/www/relate/rel_tract.html for a description of these files. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 12

share of mortgages in the state that are jumbo. Exhibit 2 Effective Interest Rates on Conventional Single-Family Mortgages 1997 1998 1999 2000 2001 2002 Georgia 7.65 7.08 7.35 8.06 6.98 6.52 North Carolina 7.58 7.00 7.24 7.99 6.96 6.52 South Carolina 7.61 7.00 7.21 8.07 6.90 6.48 Tennessee 7.65 7.03 7.35 8.11 7.06 6.64 Virginia 7.66 7.09 7.34 8.18 7.13 6.64 7.63 7.04 7.30 8.08 7.01 6.56 Between Min and Max Rate 0.08 0.09 0.14 0.19 0.23 0.16 Source: Federal Housing Finance Board, Mortgage Interest Rate Survey. However, there were important differences between other factors that also influence mortgage originations, including house prices, income, and unemployment rates (Megbolugbe and Cho, 1993), and there were larger differences between states among these factors. First, North Carolina appears to have been more severely affected by the recent recession than the other four states. Exhibit 3 shows that North Carolina s unemployment rate was the lowest of the states in this group in 1997, but the highest in 2001 and 2002. These higher-than-average unemployment rates may have affecte d mortgage activity in North Carolina. Exhibit 3 Trends in Unemployment Rates by State 1997 1998 1999 2000 2001 2002 North Carolina 3.6 3.5 3.2 3.6 5.5 6.7 Georgia 4.5 4.2 4.0 3.7 4.0 5.1 South Carolina 4.5 3.8 4.5 3.8 5.3 6.0 Tennessee 5.4 4.2 4.0 3.9 4.4 5.1 Virginia 4.0 2.9 2.8 2.2 3.4 4.1 4.6 3.8 3.8 3.4 4.3 5.1 Source: U.S. Census Bureau. North Carolina also had lower-than-average income growth during 1997-2001 with the exception of 2000 (Exhibit 4). Over the entire period, income declined by about 5 percent in North Carolina. South Carolina experienced about a 3 percent decline, while the other states all experienced a rise in income. This below-average income growth also likely slowed the growth in loan volumes in North Carolina relative to the other states. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 13

Exhibit 4 Trends in Income by State (2001 Dollars) 1997 1998 1999 2000 2001 2002 North Carolina -1.4-1.4 2.0 0.3-4.1-1.4 Georgia 10.5 4.0-0.1 5.2-3.4 10.5 South Carolina -3.2-4.3 7.6-1.9-1.0-3.2 Tennessee -2.6 9.7 5.0-10.2 2.7-2.6 Virginia 7.3-0.5 3.4 5.3-1.8 7.3 3.0 2.3 4.0-0.4-0.9 3.0 Source: U.S. Census Bureau. House price growth in North Carolina has also been below average for this group of states (Exhibit 5). In 1997, North Carolina experienced higher than average house price growth; in the years since then, growth in house prices has been below average. Only Tennessee has had relatively lower house price growth since 1999. While the fact that house prices have continued to grow faster than inflation suggests that conditions in North Carolina s housing market have been healthy, price growth has not been as rapid as elsewhere in the region. Exhibit 5 Trends in House Price by State 1997 1998 1999 2000 2001 2002 North Carolina 5.6 4.4 3.1 4.5 5.3 3.5 Georgia 5.6 6.3 5.9 6.3 7.1 4.7 South Carolina 5.1 5.5 4.6 5.1 6.2 3.9 Tennessee 5.1 5.0 2.1 3.6 5.2 3.4 Virginia 3.1 2.9 5.3 7.4 8.9 9.4 4.9 4.8 4.2 5.4 6.5 5.0 Source: Office of Federal Housing Enterprise Oversight House Price Index. In summary, North Carolina fared worse than its neighbors in unemployment and income growth in 2001 and 2002, and only Tennessee had slower house price growth. These factors would be expected to contribute to slower growth in mortgage lending. South Carolina is the most similar to North Carolina in terms of income, unemployment, and house price trends, followed by Tennessee. Georgia and Virginia both experienced stronger economic conditions for mortgage lending. As a result of the important economic differences between North Carolina and Virginia and Georgia, we limit our control group to Tennessee and South Carolina. Exhibit 6 compares Harvey and Nigro s findings with our analysis of the same four-state control group shown in Exhibit 1 and Tennessee and South Carolina only. We expect the effect of the economy to be approximately the same across North Carolina, Tennessee, and South Carolina, so differences found can be more confidently attributed to Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 14

the anti-predatory lending legislation. Note that the major patterns of changes in mortgage activity are similar across all three analyses: growth in subprime lending lagged the control group, as did growth in lending to minority and low-income borrowers. Using only Tennessee and South Carolina as the control group, however, we find essentially no difference between growth in mortgage originations by prime lenders. This result is intuitively appealing, because prime lenders are less likely to make loans with the kinds of terms affected by the legislation than subprime lenders. Exhibit 6 in Rates by Market Segment Pre - and Post-Legislation Total Originations Subprime Prime Low Income Minority Rates, Four-State Control Group North Carolina 11.1% 2.1% 12.2% -38.7% 2.5% Control Group 27.3% 36.3% 26.4% -29.3% 27.6% -16.2% -34.2% -14.1% -9.4% -25.1% Rates, Tennessee and South Carolina North Carolina 11.1% 2.1% 12.2% -38.7% 2.5% Control Group 10.7% 14.3% 10.3% -27.6% 6.0% 0.4% -12.3% 2.0% -11.0% -3.4% Harvey and Nigro -4.7% -17.2% -3.8% -3.5% -9.2% 5. Pre- and Post-Legislation Changes in Mortgage Activity North Carolina s growth in loan originations for subprime lenders after the period when the antipredatory lending legislation was implemented was substantially lower than that of control group states. This was true for both purchase and refinance loans, although there was faster growth in home improvement loans by subprime lenders in North Carolina than in control states. Changes in Mortgage Activity By Market Segment and Loan Purpose The overall growth in loan volume in North Carolina was slightly lower than growth in the control group states after implementation of the law. Exhibit 7 shows the growth in average loan volumes from 1997 and 1998 to the average in 2000, 2001, and 2002. Tables that include corresponding loan volumes pre- and post-legislation are included in the Appendix. For all loan types, North Carolina s growth was about 2.8 percentage points below the rate of growth in the control states. There were much larger differences between growth in originations among subprime lenders, where North Carolina lagged the control group by more than 16 percentage points (Exhibit 7). In comparison, among prime lenders, growth in lending volumes was roughly equal, with 12.9 percent growth in North Carolina and 13.9 percent growth in the control group. Consistent with the findings from other studies, the sizeable difference between growth in loan volumes between subprime lenders Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 15

in North Carolina and the control group indicates that the anti-predatory lending law had an effect on loan volumes for subprime lenders. in the volume of loans of all types in North Carolina also lagged growth in control group states. Overall differences were relatively small (3.9 percent among purchase loans; 3.6 percent among refinance loans; and 1.6 percent among home improvement loans). However, differences between subprime lender originations were large. in originations of purchase loans by subprime lenders was rapid in both North Carolina and the control states (110 percent growth and 137 percent growth, respectively), but faster in the control states. Among refinance loans, orig inations by subprime lenders fell by nearly 20 percent in North Carolina compared to little change in the control states, for a difference of almost 18 percentage points. Home improvement originations among subprime lenders was one of the few loan categories where growth in North Carolina exceeded the control group, and where growth in originations among subprime lenders exceeded that of prime lenders. Recall that North Carolina s anti-predatory lending law does not govern home equity loans, many of which are probably used for home improvement. It may be that subprime lenders in North Carolina have mitigated declines in refinance originations with increased focus on home equity loans. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 16

Exhibit 7 Pre- and Post-Legislation Changes in Annual Originations by Loan Purpose Percent change in average annual loan volumes from to All Loan Types Purchase Loans Refinance Loans Home Improvement Loans North Carolina Number of Loans 11 299,525 146,230 134,780 18,515 Prime 12.9% -4.2% 42.1% -32.0% Subprime 0.2% 109.8% -19.9% 12.7% Total 11.3% -0.5% 29.3% -25.2% Control Group Number of Loans 322,409 155,462 141,166 25,781 Prime 13.9% -1.1% 41.1% -24.5% Subprime 16.6% 137.3% -2.0% -16.7% Total 14.2% 3.4% 32.9% -23.5% Prime -1.0% -3.1% 1.0% -7.4% Subprime -16.4% -27.5% -17.9% 29.4% Total -2.8% -3.9% -3.6% -1.6% Changes in Mortgage Activity by Neighborhood Minority Composition In order to determine whether the effects of the anti-predatory lending legislation were disproportionately concentrated in certain types of neighborhoods, we also compared growth rates of mortgage activity in neighborhoods of varying income and minority composition. The deficit between North Carolina and the control group was largest in predominantly minority neighborhoods both overall and for prime lenders (Exhibit 8), where originations in North Carolina fell slightly, but grew by 5.2 percent in the control states. The decline in originations in these neighborhoods in North Carolina was attributable to subprime lenders, whose volume fell by 8.1 percent, compared with a small increase in loan volume among prime lenders. Among subprime lenders, the largest differences between North Carolina and the control group were in predominantly white neighborhoods, although both showed substantial increases. Here, loan 11 The number of loans in Exhibit 7 is average originations over. It is also included in Exhibit 8 to indicate that some loans were excluded from the analysis of changes in mortgage lending by the income and minority composition of the neighborhood either because the census tract number for the loan was missing or because the demographic information for the neighborhood was suppressed for confidentiality reasons. As a result of these excluded loans, overall changes in mortgage activity in Exhibit 7 and 8 are inconsistent (-2.8 percent versus 0.1 percent). Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 17

volumes among North Carolina s subprime lenders grew by about 21 percent, but loan volumes among subprime lenders in the control group grew by more than 44 percent (Exhibit 8). Exhibit 8 Pre- and Post-Legislation Changes in Loan Volume by Neighborhood Minority Composition Percent change in average annual loan volumes, to Prime Lender Subprime Lender Total North Carolina Number of Loans 282,529 250,263 32,266 >50% (predominantly minority) -1.2% 0.7% -8.1% 26-50% 11.4% 11.4% 11.8% 11-25% 21.8% 22.5% 15.5% <=10% (predominantly white) 20.7% 20.7% 21.4% Total 15.8% 16.6% 9.7% Control Group Number of Loans 311,484 278,139 33,345 >50% (predominantly minority) 5.2% 5.4% 4.6% 26-50% 13.2% 12.6% 17.8% 11-25% 17.1% 16.6% 22.4% <=10% (predominantly white) 19.1% 17.0% 44.6% Total 15.8% 14.9% 22.9% >50% (predominantly minority) -6.5% -4.8% -12.6% 26-50% -1.8% -1.2% -6.1% 11-25% 4.7% 5.9% -6.9% <=10% (predominantly white) 1.7% 3.6% -23.2% Total 0.1% 1.7% -13.2% Among home purchase loans, overall differences in growth in loan volume between North Carolina and the control states were roughly similar across neighborhood types. There were small declines in North Carolina relative to the control group in all neighborhood types except those where minorities make up 11-25 percent of the population. In these neighborhoods, there was no difference between North Carolina and the control group (Exhibit 9). There were only small differences between North Carolina and the control states among prime lenders across neighborhood types. Among subprime lenders, however, growth in home purchase loan volume was large in all neighborhood types, and lagged the control states in all areas except neighborhoods where between 26-50 percent of the population is minority. The largest lag between subprime lenders in North Carolina and the control states was in predominantly white neighborhoods, where loan volume grew about 125 percent in North Carolina, but by 186 percent in the control states. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 18

Among all refinance loans, the largest declines in North Carolina relative to the control group were in predominantly minority neighborhoods, where North Carolina s loan growth overall lagged the control states by 14.4 percentage points (Exhibit 9). This pattern is attributable to both a sizeable absolute decline in refinance lending by subprime lenders in North Carolina (22.8 percentage points) as well as much slower growth in refinance lending by prime lenders in North Carolina (15.5 percentage points) than among these lenders in the control states. Among North Carolina s subprime lenders, loan volumes fell in these neighborhoods by almost 23 percentage points compared with a drop in loan volumes in control states of about 6 percentage points. However, among subprime lenders, the largest declines in North Carolina relative to the control states were in predominantly white neighborhoods, with small growth in lending (1.3 percent) compared with 21 percent growth in the control states. Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 19

Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 20 Exhibit 9 Pre- and Post-Legislation Changes in Loan Volume by Neighborhood Minority Composition and Loan Purpose Percent change in average annual loan volumes from to All Home Purchase Loans Prime Subprime All Refinance Loans Prime Subprime North Carolina >50% (predominantly minority) -4.5% -10.9% 104.2% 7.5% 25.0% -22.8% 26-50% -0.7% -5.4% 139.4% 31.3% 43.5% -13.3% 11-25% 5.3% 1.9% 123.3% 46.1% 56.8% -8.5% <=10% (predominantly white) 2.5% -0.8% 124.6% 42.1% 49.1% 1.3% Total 1.8% -2.2% 123.9% 36.2% 48.0% -11.4% Control Group >50% (predominantly minority) -2.5% -12.1% 125.7% 21.9% 40.6% -6.4% 26-50% 1.0% -4.1% 130.0% 34.0% 44.4% -1.5% 11-25% 5.2% 1.3% 138.4% 37.1% 44.1% 0.1% <=10% (predominantly white) 5.6% 1.7% 186.1% 40.4% 43.3% 20.7% Total 3.7% -1.0% 145.8% 36.1% 43.4% 4.0% >50% (predominantly minority) -2.1% 1.2% -21.5% -14.4% -15.5% -16.4% 26-50% -1.7% -1.2% 9.4% -2.7% -0.9% -11.7% 11-25% 0.1% 0.6% -15.1% 9.0% 12.7% -8.5% <=10% (predominantly white) -3.1% -2.4% -61.5% 1.7% 5.8% -19.5% Total -1.9% -1.2% -21.9% 0.1% 4.6% -15.4%

Changes in Mortgage Activity by Neighborhood Income Composition Declines in overall mortgage lending were modest in North Carolina relative to the control group, but larger in low-income neighborhoods (by 3.3 percentage points, Exhibit 10) than in moderate- or highincome neighborhoods. The decline in lending in low-income neighborhoods was driven by an absolute decline in lending activity among subprime lenders as well as slower growth in lending among prime lenders in North Carolina. Interestingly, growth in lending in high-income neighborhoods is one of the few exceptions to the slower relative growth in North Carolina compared with the control group. Overall, average loan volume in high-income neighborhoods grew 4.6 percentage points faster in North Carolina than in the control states (Exhibit 10). Among prime lenders, lending grew 5.5 percentage points faster. Among North Carolina s subprime lenders, the largest lags were in moderate-income neighborhoods, where growth lagged the control states by 20.6 percentage points. Lags were roughly equivalent in low- and high-income neighborhoods. Exhibit 10 Pre- and Post-Legislation Changes in Mortgage Activity by Neighborhood Income Composition Percent change in average annual loan volumes from to All Prime Subprime North Carolina Low income 4.9% 6.3% -2.5% Moderate income 19.5% 19.5% 19.1% High income 26.1% 25.8% 30.4% Total 15.8% 16.6% 9.7% Control Group Low income 8.2% 7.9% 10.0% Moderate income 20.5% 18.5% 39.8% High income 21.5% 20.3% 40.6% Total 15.8% 14.9% 22.9% Low income -3.3% -1.6% -12.5% Moderate income -1.0% 1.0% -20.6% High income 4.6% 5.5% -10.2% Total 0.0% 1.7% -13.2% Among purchase loans, the largest absolute declines in loan volumes, both overall and for prime lenders, were largest in low-income neighborhoods (Exhibit 11). Among refinance loans, the largest overall lags were also in low-income neighborhoods. In contrast with very low- income neighborhoods, home purchase lending in high-income neighborhoods in North Carolina was faster Abt Associates Inc. Mortgage Lending in North Carolina After the Anti-Predatory Lending Law 21