How do predatory lending laws impact bank performance?
Hendrickson, Jill M. ; Nichols, Mark W.
INTRODUCTION
The past decade has witnessed a significant expansion of credit for
many in the United States. This is particularly true for minority
borrowers, those with limited or poor credit histories and those with
low-incomes. Banks and other financial institutions have increasingly
offered credit to this segment of the market and have frequently
benefited from higher earnings on these loans. As evidence of the growth
in this market, consider that in 1995, loan originations in the
sub-prime market were $65 billion and by 2003 originations increased to
$332 billion (Chomsisengphat and Pennington-Cross, 2006). However, as
the subprime market expanded, so did concerns regarding who was
receiving these high-priced loans. Awareness became particularly acute
in the fall of 2007 and the expansion of the financial crisis. Each
year, Housing and Urban Development (HUD) compiles a list of subprime
lenders and has used this information to make the case that much of the
subprime lending in the United States has been in low-income and
minority neighborhoods. Others, for example, Immergluck (1999) and
Marsico (2001) also find evidence that subprime lending growth is
greater in low-income and minority areas. There is also concern
regarding the outcome of these lending relationships; specifically, the
increasing number of foreclosures and loan delinquencies. Taken
together, the concern over subprime lenders targeting a certain segment
of the population and the undesirable outcome of many of these loans,
these developments prompted consumer advocacy groups and regulatory
bodies to take a look at this growing market.
The initial regulatory response was federal regulation followed
shortly by state regulation, in many states. In general, these laws
restrict high cost loans, their fees and rates. The state level
regulation varies tremendously from one state to another. Perhaps one
reason for the variation is that there is no consensus on a definition
of predatory lending. Engel and McCoy (2001) provide a set of loan
criteria that define predatory lending to include at least one of the
following: loans that provide no benefit to the borrower; loans with
misleading nondisclosures; loans with fraud or deception; loans that
require the borrower to forego redress; or loans that earn
'supranormal' profits. Pennington-Cross and Ho (2008) defined
predatory lending to be whenever the borrower is unable to understand
the terms and obligations of the loan. Morgan (2007) argues that any
welfare-reducing form of credit can be labeled predatory lending.
Similarly, the U.S. Government Accountability Office (GAO) defines a
predatory loan to be one which contains terms that will ultimately harm
the borrower (2004). Stressing the asymmetric information problem
between the borrower and lender, Morgan (2007) defines a predatory loan
to be one which the borrower would certainly decline if she shared the
same information as the lender. Clearly there is no universal
definition. However, it is accepted that predatory lending occurs within
the subprime market. Not all subprime loans are predatory but most
predatory loans are considered subprime.
Generally speaking, the predatory credit market is a subset of the
subprime market. Because of a lack of data, it is not known how large
this market actually is. There have been some micro estimates of the
size of predatory lending. For example, Goldstein (2006) estimates that
over 22 percent of all property loans in Philadelphia are predatory.
Stock (2001) investigated whether predatory lending was instrumental in
mortgage foreclosures in Montgomery County, Ohio. His sample of 1,198
mortgages indicated that 255 of these were predatory; this suggests that
approximately 21 percent of the mortgages were predatory in this region.
This limited research suggests that predatory lending is significant but
certainly more evidence is required to draw conclusions about systemic
practices.
With the first financial crisis of the twenty-first century, most
Americans became aware of the subprime market and perhaps even of
predatory lending. It seems likely that one response to the financial
crisis will be increased regulation. More specifically, it is likely
that regulators and policy makers will turn to predatory lending laws to
keep the subprime mortgage market in check. If so, it is important to
understand not only how these regulations impact the loan market but
also the institutions that are extending the loans. That is, the
predatory lending laws are, on the one hand, designed to protect
borrowers from abusive lending practices. On the other hand, however,
regulators and policy makers need to be careful that their regulatory
decisions do not end up costing the lender so much that they retreat
from this segment of the market. This paper attempts to inform
regulatory and policy discussions by empirically testing whether
predatory lending policies hurt the performance of the commercial
bankers who are extending mortgage credit.
The second section of this paper provides a concise overview of the
evolution of predatory lending laws and is followed in section three
with a review of the salient literature. This is a relatively new field
of literature as abusive lending practices have become more prevalent
with the recent national policy emphasis to increase homeownership. The
fourth section of the paper outlines the empirical specifications, data,
and methods. Section five contains the empirical results and section six
concludes. Briefly, the results indicate that national bank performance
improved relative to state bank performance following a decision by the
Office of the Comptroller to reduce the predatory lending regulatory
burden on national banks.
BRIEF BACKGROUND OF FEDERAL AND STATE PREDATORY LENDING LAWS
As credit expanded to low-income and higher-risk individuals and as
higher delinquencies and foreclosures became evident, community and
consumer advocacy groups became more vocal about their concerns that
lending abuse was taking place. The initial response to these concerns
was the passage of a 1994 federal law known as the Home Ownership and
Equity Protection Act (HOEPA) under Regulation Z at the Federal Reserve.
The purpose of HOEPA was consumer protection from potentially abusive
outcomes such as high interest rates and fees. HOEPA defined loans that
were closed-ended home equity loans that met APR and finance fee
triggers. A closed-ended loan is one in which the borrower received a
fixed sum that must be repaid over time. Protection from HOEPA was
triggered if either the loan's APR exceeded a comparable Treasury
bond by 8 percentage points on the first lien or if finance charges on
the loan exceeded 8 percent of the loan amount or a fixed $480 amount
adjusted for inflation using the consumer price index (Ho and
Pennington-Cross, 2006). Further, for HOEPA-covered loans, there were
lending restrictions which included no no-document loans, no balloon
loans, no pre-payment penalties greater than 5 years, among others. It
was anticipated that HOEPA would reduce the predatory lending that came
to light in the mid 1990s.
Nonetheless, a few years later, several government agencies
(Housing and Urban Development, the Federal Reserve Board, and the U.S.
Department of the Treasury) once again investigated the possibility of
continued abuse in this segment of the credit market. At the end of
2000, the Federal Reserve called for more stringent restrictions on
loans and an expanded definition of loans covered under HOEPA
(Elliehausen and Staten, 2004). However, regulators were also aware that
there was a fine line between protecting against lending abuses and also
servicing the credit needs of these low-income and minority credit
seekers. In the end, the changes to HOEPA recommended by the Federal
Reserve were made in 2002. Hoping to strike a balance between cutting
off credit to this segment of the market and protecting borrowers from
abusive lending practices, this further defined HOEPA-covered loans to
be those more likely to have predatory characteristics. Despite these
changes in federal law, the ongoing growth in the subprime market led
many to believe that the HOEPA changes were not sufficient. Indeed, a
2001 study found that HOEPA covers, at a maximum, only 5 percent of all
subprime mortgages (Board of Governors, 2001).
In 1999 states responded by passing their own predatory lending
laws which were more restrictive and prohibitive than the federal law.
While state coverage is much broader than HOEPA, there is significant
variance from state to state on the specifics of their predatory lending
laws. Ho and Pennington-Cross (2006) capture the extent to which state
laws extend HOEPA and also the variance between states by constructing
empirical indexes for each state with predatory lending laws. More
recently, Bostic et al. (2008) construct predatory lending law indexes
as well. However, the Bostic work attempts to refine the work of
Ho-Pennington-Cross (2006) by including four coverage measures and four
restriction measures and also considers the enforcement of the state
laws. In this way, the indexes more precisely indicate the legal
environment in each state with respect to predatory lending laws. Like
the Ho and Pennington-Cross (2006) indexes, those in Bostic et al.
(2008) provide insight to the variation of the laws themselves and the
enforcement of the laws.
The discussion above indicates that the regulatory environment for
predatory lending began with federal regulation in 1994 and was enhanced
beginning in 1999 as states began implementing even more stringent laws.
Consequently, for ten years, between 1994 and 2004, there was a mix of
both federal and state (in some states) predatory lending laws in the
United States. During this time frame, all banks operating in a state
with state predatory lending laws were required to follow these laws.
This changed in 2004 when the Office of the Comptroller of the Currency
(OCC) issued a ruling that state predatory lending laws were preempted
by federal statute (see Hawke, 2004). This meant that nationally
chartered banks, whose primary regulator is the OCC, were exempt from
state predatory lending laws. The OOC's position was that national
banks were already subject to federal anti-predatory lending laws and
also not significantly engaged in predatory lending to warrant the
burden of additional state law (Office of the Comptroller, 2004).
Thus, following the final ruling by the OCC in 2004, nationally
chartered banks were no longer legally required to meet state predatory
lending laws. This means that state chartered banks are subject to more
regulation in the post 2004 era than nationally chartered banks when it
comes to issues of abusive lending practices. Since regulation is not
without significant costs, it is reasonable to expect that nationally
chartered banks should perform better than state chartered banks in
those states with state predatory lending laws. Whalen (2008, 775)
argues that these regulations may reduce revenue if banks make fewer
mortgage loans in response to higher regulatory costs. This would,
ceteris paribus, decrease bank profits and put state chartered banks at
a competitive disadvantage to nationally chartered banks. The primary
purpose of this paper is to determine if differences in these regulatory
regimes hurts one classification of bank relative to another.
LITERATURE REVIEW
There are two salient bodies of literature that are relevant to the
purpose of this paper. First is the literature that considers the impact
of predatory lending laws on credit markets, the borrower, or the
lender. This section of the paper begins with a review of this
literature. Second, since an important assumption of this paper is that
bank regulation will increase costs, ceteris paribus, this must be
established. Consequently, the effect of bank regulation on costs is
also considered in this literature review.
It is useful to classify existing predatory literature into four
categories; that which considers the impact of predatory regulation on
the flow of credit; that which considers how the regulation impacts the
cost of credit; that which considers the likelihood of increased
foreclosures from predatory credit; and, finally, that literature that
considers the impact of the predatory regulation on the institutions
extending credit. All four are discussed separately in this section of
the paper. The first concerns the flow of subprime credit as a result of
federal and state predatory lending laws. That is, this literature asks
if the laws influenced the applications and originations for subprime
mortgages. Much of this research finds a reduction in the flow of
subprime credit (see, for example, Harvey and Nigro, 2003 and 2004, and
Elliehausen and Staten, 2004), particularly the earliest research which
focused on the North Carolina experience. Ho and Pennington-Cross (2006)
extend the initial research beyond North Carolina to include ten states
with predatory lending laws and they find that, generally speaking,
there is little to no impact on the flow of credit from this regulation.
However, their micro-analysis of the specific type of lending laws
indicates a more mixed result: some laws reduce predatory lending,
others increase it, and still others have no impact. The same authors
also find mixed results in their follow-up work on predatory lending; Ho
and Pennington-Cross (2007) find that in some states subprime
originations decrease and in others, it increases. They hypothesize that
this increase in lending may be because borrowers have fewer fears of
predation and are more willing to try and obtain a loan. Bostic et al.
(2008) also study the impact of predatory lending laws on the flow of
credit. They find that when aggregated, the laws seem to have little to
no impact of the flow of credit. However, Bostic et al. (2008) find that
particular aspects of the law may impact credit flows. More
specifically, they find that more restrictive laws reduce subprime
originations and increase the likelihood the loan will be rejected. At
the same time, they find that laws with broader coverage tend to reduce
subprime applications but also have lower rejection rates. Like Bostic
et al. (2008) and Ho and Pennington-Cross (2006), Li and Ernst (2007)
find, in the aggregate, no change in the flow of credit as a result of
predatory lending laws. Since the findings in the literature on the flow
of subprime credit are mixed, there is ongoing inquiry.
The second type of subprime literature considers the impact of the
predatory lending laws on the cost of credit. Pennington-Cross and Ho
(2008) indicate that this is more challenging than the flow literature
from the perspective that pricing in the subprime market is more complex
than in the prime market. That is, determining interest rates on
subprime loans is a function of multiple variables, for example, the
borrowers' credit score, down payment amount, specific loan
characteristics, etc. Further, increasingly, subprime loans carry
adjustable rates which, themselves, are the result of a myriad of loan
characteristics. According to Harvey and Nigro (2004), these subprime
loans often carry higher interest rates because there is a lack of
standardization in underwriting these loans which increases the cost to
both originate and service the loans. That said, Pennington-Cross and Ho
(2008) hypothesize that states with predatory lending laws have higher,
systematic, interest rates on subprime mortgage loans as a result of
higher regulatory compliance costs. To test their hypothesis, the
authors perform two estimations with two different data sets. In the
first, they use Home Mortgage Disclosure Data (HMDA) to examine the
impact of lending laws on the annual percentage rates (APRs) of a
subprime loan. The find no empirical evidence that predatory lending
laws lead to higher APRs. In the second estimation, they rely on
interest rate data from LoanPerformance Inc. on securitized subprime
loans which is a data set largely confined to the top segment of the
subprime population, known as the A- segment. They find a small increase
in interest rate costs for fixed rate loans and a small decrease in
interest rate costs for adjustable rate loans.
Many states that adopted predatory lending laws did so to reduce
the number of foreclosures that are said to be the result, in part, from
predatory loans. The third classification of predatory regulation
literature investigates how these laws impact foreclosures. For example,
Rose (2008) investigates the Chicago metropolitan area and determines
that lending laws may or may not be effective in influencing foreclosure
rates. His empirical analysis suggests that the relationship between
foreclosure rates and loan characteristics is complicated; not all
limitations on lending found in predatory lending laws reduce
foreclosures. Quercia et al. (2007) ask if loans with predatory
characteristics are associated with an increased likelihood of
foreclosure. These authors find significant evidence to suggest that
refinanced loans with specific predatory loan characteristics are more
likely to foreclose than other types of non-predatory loans. Generally,
these findings have implications for the regulation of predatory, and
other, type of lending.
A fourth subset of the predatory lending literature concerns the
impact that predatory lending laws have on bank performance as a result
of regulatory compliance. For example, Whalen (2008) uses an event study
approach to determine if the OCC's preemption decisions in Georgia
that allow national banks to circumvent state predatory lending laws
give an advantage to one charter type over another. Using stock values
to measure wealth, Whalen (2008) finds that the preemption did not
create an advantage for national banks; only small, geographically
diverse national banks witnessed increased performance. He offers the
possibility that preemption reduced these small bank costs attributable
to regulatory compliance and scale. Whalen (2008) is the only work that
considers the impact of predatory regulation on the lending
institutions. Consequently, there is room for further analysis on the
issue of the impact of predatory regulation on the institutions affected
by the laws.
A comprehensive review of the regulatory costs in banking is found
in Elliehausen (1998). In this work, Elliehausen carefully documents the
research related to methods of measuring regulatory costs including case
studies, surveys, and econometric analysis, among others. Across all
methodologies, the findings are consistent: regulation increases the
cost of delivering banking products and services. Further, compliance
costs are significantly greater for smaller banks. That is, there are
scale economies for regulatory costs in banking.
Given the overwhelming evidence found in Elliehausen (1998), it
seems safe to assume that regulation increases bank costs. Consequently,
it is reasonable to ask if predatory lending regulation laws put state
bank performance at a disadvantage relative to national banks. In doing
so, this paper makes several important contributions to the literature.
First, to the author's knowledge, there is no existing work,
outside of Whalen (2008), that investigate how predatory lending laws
impact bank performance. If banks pursue lending strategies to maximize
their profits and if laws are created to change the nature of lending,
it seems this would negatively impact bank profitability and stability.
Whalen (2008) admits that there is no empirical evidence related to
these issues so that the debate about both predatory lending and the
preemption's effects on banks is ongoing. Where Whalen (2008)
considered the impact at bank holding companies on their wealth (stock
value), this paper considers several measures of bank performance using
balance sheet and income statement data.
Second, this paper relies on some unique data that has not been
utilized in the literature to date. The three most common sources of
data in the predatory lending literature are the HMDA data used by
Pennington-Cross and Ho (2008), Ho and Pennington-Cross (2006 and 2008),
Bostic et al. (2008) and Whalen (2008), the HUD data used in Harvey and
Nigro (2004), Ho and Pennington-Cross (2006) and Bostic et al. (2008)
and the LoanPerformance Inc. data (a private data vendor) in
Pennington-Cross and Ho (2008), Li and Ernst (2007), Quercia et al.
(2007), Demyanyk and Hermert (2008), and Rose (2008). A less utilized
data source in the literature is the American Financial Services
Association data in Elliehausen and Nigro (2004). This paper follows
existing literature by utilizing the HMDA Aggregate Reports data to
identify the banks in our sample and then relies on the FDIC for firm
level data. However, this paper also utilizes survey data from the
Conference of State Bank Supervisors to identify the legal bank
environment in each state.
IDENTIFICATION AND ESTIMATION
The hypothesis tested in this paper is that the OCC's
preemption of state predatory lending laws improves the performance and
stability of nationally chartered banks relative to state charted banks
in those states with predatory lending laws. To empirically test this
hypothesis, a difference-in-difference-in-differences (DDD) model is
estimated. Before presenting this, however, the sample of states with
and without predatory lending laws is described.
Identifying which states have predatory lending laws is not as
straightforward as one may expect because there are different
definitions of predatory lending laws. There are two primary existing
works that attempt to quantify state predatory lending laws; these are
Ho and Pennington-Cross (2006) and Bostic et al. (2008). Ho and
Pennington-Cross (2006) construct three indexes for each state; one
captures the extent to which the state law broadens HOEPA; a second
captures how restrictive each state is relative to others; and the third
combines the first two for an overall measure of regulatory rigor. The
difference in these indexes highlights the variation across states and
also provides evidence for identifying which states have a greater
regulatory burden. In their work, Ho and Pennington-Cross (2006) focus
on states with laws that use triggers similar to HOEPA to define an
abusive loan. Consequently, some states with predatory lending laws, but
without triggers, are not in their sample.
Bostic et al. (2008) also construct predatory lending indexes for
each state. However, these authors differentiate between anti-predatory
lending laws and what they call "mini-HOEPA" laws. Mini-HOEPA
laws are typically triggered below the national HOEPA triggers, but
state differences remain in terms of enforcement and restrictions.
Predatory lending laws are, to these authors, statutes or regulation of
prepayment penalties, balloon clauses or other loan characteristics that
are often considered predatory. Using their definition of both predatory
lending laws and mini-HOEPA laws, Bostic et al. (2008) determine that by
January 1, 2007, only six states (Arizona, Delaware, Montana, North
Dakota, Oregon, and South Dakota) had no predatory lending laws or
mini-HOEPA laws in place. Because Ho and Pennington-Cross (2006) use a
sample of banks and because Bostic et al. (2008) use a broader
definition of predatory lending laws, these two works do not identify
the same states in the creation of their indexes.
To complicate matters, the "Profile of State Chartered
Banking" (2008) at the Conference of State Bank Supervisors (CSBS),
surveys state banking authorities on a wide range of budget, cost, and
regulatory issues. The survey in the most recent profile (2008), asks
the state authorities if there are predatory lending laws in their state
during the 2006-2007 period. Alabama, Alaska, Arizona, Delaware, Hawaii,
Indiana, Iowa, Louisiana, Michigan, Missouri, Montana, Nebraska, New
Hampshire, North Dakota, Oregon, South Dakota, Texas, Utah, Virginia,
Washington, and Wyoming all indicate that there are no predatory lending
laws in place. Because of inconsistencies in these three data sources,
the authors contacted directly the state bank authorities for
clarification. These conversations relieved that Texas, Michigan,
Indiana, and Utah do have predatory lending laws. Consequently, we
identify seventeen states without predatory lending laws using the
survey data.
Since this list of seventeen is generally consistent with Ho and
Pennington-Cross (2006), we use this as our sample of states without
predatory lending laws. Table 1 lists all states; those with shading are
identified by the "Profile of State Chartered Banking" (2008)
and confirmed by author interviews as not having predatory lending laws
and those listings without shading do have predatory lending laws. The
effective date of the law is identified in the second column. The final
two columns indicate the index calculated by Ho and Pennington-Cross
(2006) and by Bostic et al. (2008). In both cases, a higher number
indicates more restrictive lending laws.
Since we are interested in learning how the preemption by the OCC
of state predatory lending impacts national bank performance relative to
state bank performance, our sample requires banks operating under state
predatory lending laws and a control sample of banks not operating under
state predatory lending laws. While one option is to simply compare bank
performance at the state level between the different legal environments,
it may be more instructive to consider bank performance at a more micro
level. To that end, this paper uses metropolitan statistical area (MSA)
level data to compare bank performance, by charter type, in MSAs with
and without predatory lending laws. The process of selecting MSAs relies
on first identifying the largest six MSAs in each legal environment:
those states with and those without predatory lending laws. The U.S.
Census Bureau identifies the MSAs by population rank and, as of July 1,
2008, table 2 identifies the largest six MSAs for those states with and
without predatory lending laws.
Recall that Bostic et al. (2008) claim that only six states, at the
beginning of 2007, do not have predatory lending laws. This is because
Bostic et al. use a different definition of predatory lending than do
some state authorities. To test the robustness of our sample and
results, we also collect data on the largest MSA in each of the six
states identified in Bostic et al. (2008) and we will apply the same
estimation techniques to this sample as to the sample identified under
the Ho and Pennington-Cross (2006) and "Profile of State Chartered
Banking" (2008) procedure explained above. Table 2 also identifies
the Bostic et al. (2008) sample.
Once the MSAs are identified, the institutional directory at the
FDIC identifies all commercial banks headquartered in each MSA. Several
of the MSAs in our sample include more than one state with conflicting
predatory lending laws. In these cases, for example, the Fargo MSA
includes parts of Minnesota, the banks headquartered in states with
inconsistent predatory lending laws are removed from the sample (e.g.
Minnesota banks who operate under predatory lending laws are removed
from this MSA). Home Mortgage Disclosure Act (HMDA) Aggregate Reports
identify all banks and financial institutions headquartered in each MSA.
Using this list, all banks are cross-listed with the FDIC's
Institutional Directory to determine if the bank is a commercial bank
and to determine its charter type. We end up with three samples that
will be used to test the hypothesis that predatory lending laws impact
bank performance. The sample extends from 1999 through 2008.
To capture bank performance, we use three variables identified by
Nippani and Green (2002): 1) real return on equity (ROE); 2) real return
on assets (ROA); and 3) real net interest margin (NIM). These are
represented as PERFORMANCE in equation (1) below. Data definitions and
sources are found in the Data Appendix. The DDD model estimated in our
analysis is as follows:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (1)
[PLL.sub.i,j,t] is a dummy variable equal to one if bank i in MSA
j, at time t, is subject to predatory lending laws, zero otherwise.
[OCC.sub.t] is a dummy equal to one for years 2004 through 2009, and
captures the decision by the Office of the Comptroller to preempt state
predatory lending laws for all nationally chartered banks, effective
January 7, 2004 (Whalen, 2008). [NAT.sub.i,j,t] is a dummy variable
equal to one if bank i in MSA j at time t is a nationally chartered
bank, zero otherwise.
The various interaction terms relate to the DDD model. Intuitively,
a difference-indifference (DD) model compares two different groups over
different time periods. Frequently one group is subject to an
intervention or treatment in one period(s), but not the other period(s).
Another group is not subject to the intervention in either period(s).
Rather than simply comparing the treatment group before and after the
intervention, the DD model nets out, or controls for, changes in the
non-treatment group that also occurred over the different periods. In
our case, the intervention is the decision by the Comptroller to exempt
national banks for state predatory lending laws. However, what is the
appropriate control group? One DD model could compare national bank
performance with state bank performance in states with predatory lending
laws. Alternatively, another DD model could compare national bank
performance in states with predatory lending laws to national bank
performance in states without predatory lending laws. The DDD model
above is a more robust model that uses both state banks in predatory
lending law states and national banks in non-predatory lending law
states as controls. The coefficient, [[beta].sub.7], on the interaction
term NAT*OCC*PLL, provides the additional return to national banks after
the Office of Comptroller's decision in predatory lending law
states (for other examples of the difference-in-difference-indifferences
estimator see Wooldridge, 2007, Currie et al., 2009, and Acs and Nelson,
2004).
Other variables in equation (1) are included as various controls.
[STATE.sub.j] is a dummy variable for the state that MSA j is located
in. [YEAR.sub.t] is a set of year dummy variables. [ACTIVITY.sub.i,j,t]
represents bank activity that can impact bank performance. Specific
variables included in ACTIVITY are: the real total amount of 1-4 family
home loans extended by the bank (ML) in hundreds of thousands of
dollars. During healthy times, it is expected that ML contributes
positively to bank performance. However, during times of falling home
values, it is expected that ML hurts bank performance. The ratio of 1-4
family mortgage loans made to low-income applicants to total 1-4 family
mortgage loans extended (LIM) is also included. If low-income loan
recipients are at a higher risk of default, these loans may hurt bank
performance. The bank's real security investments (SEC), in
hundreds of thousands of dollars, are also included and since banks are
limited to government securities, these investments are low risk and
should enhance, or at least not hurt, bank performance. Finally,
banks' most important source of fund is deposits. Consequently,
this is an important cost on their balance sheets. To control for these
costs, this paper relies on the real interest deposit expenses, IDE, at
each bank in the sample, also measured in hundreds of thousands of
dollars. It is expected to negatively impact bank performance since it
represents a cost to the bank.
Finally, equation (1) also contains several controls for the local
market that each bank operates within since the market structure
certainly impacts bank performance: the number of financial institutions
(NUM) either headquartered in the MSA or with branches in the MSA
captures the competitive nature of the MSA. Real per capita personal
income, PCPI, in thousands of dollars, describes the borrowing
population facing the banks. Finally, the percentage change in home
values (HV) within the MSA is included. Certainly, a declining house
market will hurt bank performance since the potential for loan defaults
increases. The regression specifications in equation (1) also control
for general movements in interest rates by including the prime rate,
PRIME.
RESULTS
Results from estimating equation (1) on our three samples are
presented in Tables 3-5. Table 3 provides estimates using the Return on
Assets (ROA) as the performance measure. The full sample, reported in
the first column of results, consists of banks located in all the MSAs
identified in Table 2. Thus the full sample includes states identified
by both Ho Pennington-Cross (2006) and Bostic (2008) as not having
predatory lending laws and, of course, banks located in MSAs in states
with predatory lending laws. The second column of results is based on
the sample where states without predatory lending laws are those
identified by Ho Pennington-Cross (2006) and the Profile of State
Chartered Banking. Finally, the third column is based on the sample
where states without predatory lending laws are limited to those
identified by Bostic (2008). Tables 4 and 5 present results for the
return on equity and net interest margin.
Our main estimated coefficient of interest, [[beta].sub.7], is on
the interaction term NAT*OCC*PLL. As shown in Table 3, national banks in
states with predatory lending laws had a significant increase in their
return on assets after the Office of the Comptroller's ruling
exempting them from a state's predatory lending laws. This is true
regardless of the sample that is examined, i.e., regardless of the
definition used to identify states without predatory lending laws. The
average return on assets for the whole sample is 0.678, with a standard
deviation of 5.24, so the increase to national banks of 1.793 is
economically meaningful as well.
It is important to place this increase in the return on assets in
context and recognize that it is the marginal change in the return on
assets for those banks in predatory lending states after the OCC
decision. For example, Table 3 demonstrates that national banks had
higher returns, ceteris paribus, than state banks irrespective of
location or time, demonstrated by the coefficient, [[beta].sub.3], on
NAT. Similarly, Table 3 also demonstrates that all banks performed more
poorly after the Office of the Comptroller's ruling (2004),
regardless of charter type, as seen by [[beta].sub.2], the coefficient
on OCC. This is not surprising since this covers the 2004-2008 period
and bank profits were compromised as a part of the wider financial
crises during this period.
The other interactions also show marginal effects for various
place-time-charter type groupings. For example, the coefficient,
[[beta].sub.4], on NAT*PLL indicates that national banks in those states
with predatory lending laws had a decrease in their profitability
relative to their national bank counterparts that were not subject to
predatory lending laws. Combined with the coefficient on NAT*OCC*PLL,
this suggests that the predatory lending laws compromised profitability
for national banks in general, but that this reduction was offset after
the Office of the Comptroller's exemption. This is consistent with
the findings in Elliehausen (1998), who finds regulation increases bank
costs and hence reduces profit. This finding is also consistent with the
Comptroller's concern that state predatory lending laws
unnecessarily burden national banks. Similarly, the coefficient,
[[beta].sub.5], on NAT*OCC indicates that national bank profit was
declining after 2004 relative to earlier years, regardless of location.
The above is best demonstrated with a few examples. Consider the
marginal impact on the return on assets for national banks in predatory
lending law states after the Comptroller's decision. In this case,
NAT=1, PLL=1, and OCC=1. Ceteris paribus, the marginal impact on the
return on assets is 0.207, calculated by summing the corresponding
coefficients (0.655-0.803 + 0.980-1.162-1.373 + 0.117 + 1.793).
Comparing this for state banks in predatory lending law states during
the same period (NAT=0, PLL=1, OCC=1) yields a marginal impact of -0.03
(0.655-0.803 + 0.117). Similarly, the marginal impact on the return on
assets for national banks in non-predatory lending law states after the
Comptroller's decision (NAT=1, PLL=0, OCC=1) is -1.19 (-0.803 +
0.980-1.373). Even if one excludes the coefficient on OCC*PLL (0.117)
due to it being statistically insignificant the qualitative conclusions
do not change: the national bank's exemption from state predatory
lending laws reduced the negative impact of the financial crisis and
improved their profitability compared to their state-chartered
counterparts and national bank counterparts in non-predatory lending law
states.
Other variables in the DDD regression are also statistically
significant. The results in Table 3 reveal, for example, higher real
security holdings increase a bank's ROA as does an increase in home
values. Most of the other bank activity control variables are not
statistically significant.
Tables 4 and 5 present similar conclusions when examining the
bank's real return on equity (ROE) and net interest margin (NIM).
While not as statistically strong, Table 4 shows a significant increase
in a national bank's ROE in predatory lending law states after the
Office of Comptroller's decision. The only exception is when
Bostic's (2008) sample of non-predatory lending law states is used
for comparison. The increase, of 7.975 and 9.409 for the full and Ho and
Pennington-Cross (2006) samples respectively, is notable given the
average ROE of 8.00 with a standard deviation of 57.02. When examining
NIM, a similar conclusion is reached, only this time it is the Bostic et
al. (2008) sample that is statistically significant. The average NIM is
3.76 with a standard deviation of 1.957.
Recall that both Ho and Pennington Cross (2006) and Bostic et al.
(2008) construct predatory lending law indices that differentiate states
by the severity of their predatory lending laws. Including these indices
in the above regressions yielded a negative and statistically
significant impact on ROA, further suggesting that predatory lending
laws harm bank performance. However, the indices were not statistically
significant in the regressions on ROE and NIM.
In summary, the above results suggest the national banks in states
with predatory lending laws benefited from the decision to exempt them
from those laws, both relative to national banks in states without
predatory lending laws and compared to state banks in states with
predatory lending laws.
CONCLUSION
The only existing research that investigates the impact of
predatory lending laws on bank performance is Whalen (2008) who finds
that the Office of the Comptroller's preemption decision did not
significantly increase stock prices of national banks. This study, which
uses a much broader sample of banks and different measures of bank
worth, finds national bank performance improved after the
Comptroller's ruling when controlling for charter type and legal
environment. While not directly comparable to Whalen due to the
different measures of bank worth, these results do support his argument
that regulation may reduce revenue as well as the hypothesis in Ho and
Pennington-Cross (2007) that there may be a regulatory cost burden
associated with predatory lending laws. While we are unable to
distinguish whether it is reduced revenue or increased costs that harm
performance, it is clear that nationally chartered banks in predatory
lending law states outperform their state bank counterparts and their
national bank counterparts in non predatory lending states after the
Comptroller's ruling.
Going back to the early 1990s, the federal government has been
pursuing a set of policies to increase access to homeownership. An
unintended consequence of these policies has been the increased
incidence of predatory lending practices. The results of this paper
suggest that policymakers should be careful when trying to address
predatory lending issues. Costly regulation is likely to hurt bank
performance and poor performing banks will not be able to extend credit,
mortgage or otherwise, to that portion of the population targeted
through federal policy. If regulation is too burdensome and too costly,
it is possible that banks will respond by either increasing costs to
borrowers or restricting credit to higher risk borrowers. Neither
response by banks will help achieve the wider homeownership goals that
have characterized federal policy for the past twenty years.
Data Appendix: Data Definitions and Sources
Variable Description Source
Dependent Variables
Real Return on Ratio of undivided FDIC.gov
Equity (ROE) profits to total
assets at commercial
banks in the sample.
Real Return on Ratio of undivided FDIC.gov
Assets (ROA) profits to equity
capital at
commercial banks in
the sample.
Real Net Interest Ratio of difference FDIC.gov
Margin (NIM) in interest income
and interest expense
to earning assets at
commercial banks in
the sample.
Control for Local
Market Conditions
Number of Financial otal number of FDIC.gov
Institutions (NUM) commercial banks
headquartered in the
MSA.
Per Capita Personal Per capita personal BEA/gov/regional/
Income (PCPI) income in the MSA. reis/
House Values (HV) Index of house FHFA.gov
prices in the MSA.
Control for Legal
Environment
Binary Control (PLL) simple control for o and Cross-
states with and Pennington (2006),
without predatory Bostic et al.
lending laws = 0 if (2008), and the
no laws and = 1 if rofile of State
the states does have Chartered Banking
laws according to (2008).
the date the law was
implemented.
OCC Decision (OCC) A control for the N.A.
OCCs decision to
preempt state
predatory lending
laws= 0 for 1999-
2003 and = 1 for
2004-2009.
Bank and Borrower
Characteristics
Home Mortgages to Ratio of mortgage HMDA Disclosure
Low-Income (LIM) loans to low-income Reports
applicants to total
mortgages loans by
the commercial banks
in the sample.
Home Mortgage Loans Dollar value of 1-4 FDIC.gov
(ML) family home loans
extended by the
commercial banks in
the sample.
Securities Dollar value of FDIC.gov
Investments (SEC) securities
investments by the
commercial banks in
the sample.
Interest Deposit Dollar value of FDIC.gov
Expense (IDE) interest expense on
deposits by the
commercial banks in
the sample.
Other
CPI Consumer price index BLS.gov
used to convert
nominal to real
values.
Prime Rate (PRIME) A general control stlouisfed.org
for interest rate .fred2.data.PRI
movements. ME.txt
REFERENCES
Acs, Gregory and Nelson, Sandi (2004). Changes in Living
Arrangements During the Late 1990s: Do Welfare Policies Matter. Journal
of Policy Analysis and Management, 23(2), 273-290.
Board of Governors (2001). Truth in Lending. 66 Fed. Reg. 65,604.
Bostic, Raphael W., Kathleen C. Engel, Patricia A. McCoy, Anthony
Pennington-Cross, and Susan M. Wachter (2008).
State and Local Anti-Predatory Lending Laws: The Effect of Legal
Enforcement Mechanisms. Journal of Economics and Business, 60, 47-66.
Conference of State Bank Supervisors (2008). Profile of State
Chartered Banking.
Chomsisengphat, S., and Anthony Pennington-Cross (2006). The
Evolution of the Subprime Mortgage Market. Federal Reserve of St. Louis
Economic Review, 88, 31-56.
Currie, Janet; Hanushek, Eric; Khan, Megan; Neidell, Matthew; and
Rivkin, Steven (2009). Does Pollution Increase School Absences. The
Review of Economics and Statistics, November 2009, 91(4), 682-694.
Demtanyk, Yuliya, and Otto Van Hemert (2008). Understanding the
Subprime Mortgage Crisis. Retrieved October 14, 2009 from
http://ssrn.com/abstract=1020396
Elliehausen, Gregory (1998). The Cost of Banking Regulation: A
Review of the Evidence. Board of Governors of the Federal Reserve System
Staff Study, 171.
Elliehausen, Gregory and Michael E. Staten (2004). Regulation of
Subprime Mortgage Products: An Analysis of North Carolina's
Predatory Lending Laws. The Journal of Real Estate Finance and
Economics, 30, 167-196.
Engle, Kathleen, and Patricia McCoy (2001). The Law and Economics
of Remedies of Predatory Lending. Proceedings of a Federal Reserve
System Community Affairs Research Conference: Changing Financial Markets
and Community Development, April 5-6.
General Accountability Office (2004). Consumer Protection: Federal
and State Agencies Face Challenges in Combating Predatory Lending,
GAO-04-280.
Goldstein, Ira (2006). Predatory Lending and Mortgage Foreclosures
in Philadelphia. Retrieved September 24, 2009 from
http:www.trfund.com/resource/downloads/staffwritings/ABA_NLADA_2006.pdf
Harvey, Keith D. and Peter J. Nigro (2003). How Do Predatory
Lending Laws Influence Mortgage Lending in Urban Areas? A Tale of Two
Cities. Journal of Real Estate Research, 25, 479-508.
Harvey, Keith D. and Peter J. Nigro (2003). Do Predatory Lending
Laws Influence Mortgage Lending? An Analysis of the North Carolina
Predatory Lending Law. Journal of Real Estate Finance and Economics,
29(4), 435-456.
Hawke, John D. (2004). Statement of Comptroller of the Currency
John D. Hawke, Jr. Regarding the Issuance of Regulations Concerning
Preemption and Visitorial Powers. Retrieved September 21, 2009 from
http://www.occ.treas.gov/2004-3aComptrollersstatement.pdf
Ho, Giang and Anthony Pennington-Cross (2006). The Impact of Local
Predatory Lending Laws on the Flow of Subprime Credit. Journal of Urban
Economics, 60(2), 210-228.
Ho, Giang, and Anthony Pennington-Cross (2007). The Varying Effects
of Predatory Lending Laws on High-Cost Mortgage Applications. Federal
Reserve Bank of St. Louis Review, January/February, 39-59.
Immergluck, Daniel (1991). Two Steps Back: The Dual Mortgage
Market, Predatory Lending, and the Undoing of Community Development.
Chicago: The Woodstock Institute.
Kaper, Stacy (2007). Bernanke: Predator Bill Would Be a Good Idea.
American Banker, 172(3), 1. Retrieved September 24, 2009 from
http:www.americanbanker.com.
Li, Wei, and Keith S. Ernst (2007). Do State Predatory Lending Laws
Work? A Panel Analysis of Market Reforms. Housing Policy Debate, 18(2),
347-391.
Marsico, Richard D. (2001). Patterns of Lending to Low-Income and
Minority Persons and Neighborhoods: The 1999 New York Metropolitan Area
Lending Scorecard. New York Law School, Public Law Research Paper 02/02.
Morgan, Donald P. (2007). Defining and Detecting Predatory Lending.
Federal Reserve Bank of New York Staff Report No. 273. Retrieved
September 28, 2009 from
http://www.newyorkfed.org/research/staff_reports/sr273.pdf
Office of the Comptroller (2004). OCC Issues Final Rules on
National Bank Preemption and Visitorial Powers; Includes Strong Standard
to Keep Predatory Lending out of National Banks. News Release. Retrieved
September 22, 2009 from
www.occ.treas.gov/toolkit/newsrelease.aspx?doc=zn9i8h7t.xml
Pennington-Cross, Anthony, and Giang Ho (2008). Predatory Lending
Laws and the Cost of Credit. Real Estate Economics, 36, 175-211.
Quercia, Roberto G., Michael A. Stegman, and Walter R. Davis
(2007). The Impact of Predatory Lending Loan Terms on Subprime
Foreclosures: The Special Case of Prepayment Penalties and Balloon
Payments. Housing Policy Debate, 18(2), 311-346.
Renuart, Elizabeth (2004). An Overview of the Predatory Mortgage
Lending Process. Housing Policy Debate, 15(3), 467-502.
Rose, Morgan J. (2008). Predatory Lending Practices and Subprime
Foreclosures: Distinguishing Impacts by Loan Category. Journal of
Economics and Business, 60, 13-32.
Stock, Richard (2001). Predation in the Sub-Prime Lending Market:
Montgomery County. Center for Business and Economic Research. Retrieved
September 24, 2009 from
www.mvfairhousing.org/cber/pdf/Executive%20summary.PDF
Whalen, Gary W. (2008). The Impact of Preemption of the Georgia
Fair Lending Act by the OCC on National and State Banks and the Dual
Banking System. The Quarterly Review of Economics and Finance, 48,
772-791.
Woodridge, Jeff (2007). What's New in Econometrics, Lecture
10, Difference in Difference Estimation. Retrieved January 10, 2010 from
http://www.nber.org/WNE/Slides7-31-07/slides_10_diffindiffs.pdf
Jill M. Hendrickson, University of St. Thomas
Mark W. Nichols, University of Nevada
Table 1. Predatory Lending Laws and Legal Indices
Full State Index Full State
States Effective Year from Ho and Index from
of Predatory Pennington- Bostic et
Lending Law Cross (2006) al. (2008)
Alaska
Alabama
Arizona
Arkansas 7/16/2003 8 6.56
California 7/01/2002 11 4.93
Colorado 6/07/2002 13 4.18
Connecticut 10/02/2001 10 4.88
Delaware
Florida 10/2/2002 4 3.75
Georgia 10/2/2002 16 6.83
Hawaii
Idaho 7/01/2003
Illinois 1/01/2004 13 8.11
Indiana 2/04/2005 11 6.76
Iowa
Kansas 4/04/1999
Kentucky 6/24/2003 9 5.86
Louisiana
Maine 9/13/2003 4 3.01
Maryland 5/16/2002 8 3.39
Massachusetts 3/22/2001 14 8.44
Michigan 12/23/02 5.99
Minnesota 8/01/2007 7.01
Mississippi 7/01/2002
Missouri
Montana
Nebraska
Nevada 10/01/2003 4 2.81
New Hampshire
New Jersey 11/27/2003 10.5 7.34
New Mexico 1/01/2004 17 9.90
New York 4/01/2003 10 5.82
North Carolina 10/01/1999 11 6.40
North Dakota
Ohio 2/22/2002 6 3.47
Oklahoma 1/01/2004 9 4.29
Oregon
Pennsylvania 6/21/2001 7 3.47
Rhode Island 12/31/2006
South Carolina 1/01/2004 9 4.80
South Dakota
Tennessee 1/01/2007
Texas 9/01/2001 8 4.34
Utah 9/01/2004 6 3.91
Vermont 9/16/1998
Virginia
Washington
West Virginia 6/4/2002 9.00
Wisconsin 2/01/2005 5
Wyoming
Source: State Profile of State Chartered Banking (2008) and author
interviews with state bank authorities to determine if there is
regulation in place (all non-shaded states) and for the effective date;
Ho and Pennington-Cross (2006) for their full index which captures the
law at the end of 2004, and Bostic et al. (2008) for their full,
additive index for 2004-2005.
Table 2. Largest MSAs in States with and without Predatory Lending Laws
6 Largest MSAs in 6 Largest MSAs in Largest MSAs in
states with States without Bostic Definition of
Predatory Lending Predatory Lending non Predatory
Laws Laws Lending Law States
New York-Northern Phoenix-Mesa- Phoenix-Mesa-
New Jersey-Long Scottsdale Scottsdale
Island
Los Angeles-Long Seattle-Tacoma- Sioux Falls
Beach-Santa Ana Bellevue
Chicago-Naperville- St. Louis Dover
Joliet
Dallas-Fort Portland-Vancouver- Portland-Vancouver-
Worth-Arlington Beaverton Beaverton
Philadelphia- Kansas City Fargo
Camden-Wilmington
Houston-Sugarland- Richmond Billings
Baytown
Table 3. DDD Model for the Impact of Predatory Lending Laws on
National Bank Performance: Real Return on Assets
Profile of
State Chartered
Banking (2008)
Full Sample and Ho and Bostic et al.
Pennington- (2008) Sample
Cross (2006) Sample
PLL 0.655 *** 0.637 *** 0.758 ***
(2.76) (2.70) (2.66)
OCC -0.803 ** -1.275 *** -2.175 ***
(2.47) (5.01) (2.95)
NAT 0.980 *** 0.866 *** 1.045 ***
(3.19) (2.79) (3.24)
NAT*PLL -1.162 *** -1.046 *** -1.235 ***
(3.41) (3.05) (3.47)
NAT*OCC -1.373 ** -1.795 *** -1.118 *
(2.49) (2.90) (1.78)
OCC*PLL 0.117 0.153 1.645 **
(0.54) (0.64) (2.55)
NAT*OCC*PLL 1.793 *** 2.218 *** 1.538 **
(3.09) (3.43) (2.35)
ML -0.0006 0.0017 0.00001
(0.87) (1.23) (0.02)
LIM -1.109 4.357 -17.500
(0.12) (0.43) (1.28)
SEC 0.0011 * 0.0085 0.0014 **
(1.65) (1.32) (1.99)
IDE -0.0003 -0.0031 -0.0053
(0.08) (0.71) (1.16)
NUM 0.002 0.002 0.0031
(0.74) (0.77) (1.07)
HV 0.024 *** 0.025 *** 0.0326 ***
(3.22) (3.47) (4.21)
PCPI 0.036 0.001 0.0074
(0.77) (0.03) (1.30)
PRIME 0.068 0.101 0.0461
(0.92) (1.33) (0.46)
State Dummies Yes Yes Yes
Year Dummies Yes Yes Yes
N 8277 7922 6666
[R.sup.2] 0.030 0.027 0.035
Absolute value of the t statistic in parentheses. A *, **, and ***
represent statistical significance at the 10, 5, and 1 percent level,
respectively.
Table 4. DDD Model for the Impact of Predatory Lending Laws on
National Bank Performance: Real Return on Equity
Full Sample Profile of Bostic et
State Chartered al. (2008)
Banking (2008) Sample
and Ho and
Pennington-Cross
(2006) Sample
PLL -1.233 -1.251 -0.806
(0.42) (0.46) (0.28)
OCC -4.657 ** -11.182 *** -6.714
(2.34) (8.78) (1.60)
NAT 6.592 6.459 7.572
(1.60) (1.55) (1.62)
NAT*PLL -7.752 * -7.611 -8.730 *
(1.64) (1.58) (1.66)
NAT*OCC -6.795 * -8.227 ** -6.407
(1.68) (2.15) (1.36)
OCC*PLL -1.368 -1.144 0.054
(0.76) (0.61) (0.01)
NAT*OCC*PLL 7.975 * 9.409 ** 7.575
(1.92) (2.41) (1.58)
ML -0.001 0.007 -0.0004
(0.20) (0.77) (0.07)
LIM 23.095 19.846 43.076
(0.34) (0.50) (0.42)
SEC 0.014 ** 0.015 ** 0.015 **
(2.10) (2.23) (2.20)
IDE 0.0054 -0.021 -0.173
(0.11) (0.43) (0.36)
NUM -0.078 -0.077 -0.084
(0.76) (0.75) (0.76)
HV 0.105 0.127 * 0.093
(1.40) (1.72) (0.98)
PCPI 0.159 0.141 0.171
(1.32) (1.13) (1.40)
PRIME 0.644 0.371 0.724
(1.444) (1.25) (1.29)
State Dummies Yes Yes Yes
Year Dummies Yes Yes Yes
N 8277 7922 6666
[R.sup.2] .010 .010 .010
Absolute value of the t statistic in parentheses. A , **, and
** represent statistical significance at the 10, 5, and 1
percent level, respectively.
Table 5. DDD Model for the Impact of Predatory Lending Laws on National
Bank Performance: Real Net Interest Margin
Full Sample Profile of Bostic et
State Chartered al. (2008)
Banking (2008) Sample
and Ho and
Pennington-Cross
(2006) Sample
PLL -0.119 -0.189 ** -0.084
(1.12) (2.21) (0.69)
OCC -1.328 *** -1.000 *** -1.667 ***
(6.52) (10.68) (6.60)
NAT 0.167 0.097 0.173
(1.55) (1.57) (1.49)
NAT*PLL -0.154 -0.089 -0.161
(1.05) (0.75) (1.06)
NAT*OCC -0.380 ** -0.177 -0.415 *
(1.99) (1.37) (1.63)
OCC*PLL 0.218 ** 0.0929 0.699 ***
(1.96) (0.96) (2.68)
NAT*OCC*PLL 0.415 * 0.212 0.454 *
(1.87) (1.25) (1.63)
ML -0.0027 *** -0.0063 *** -0.0023 ***
(4.04) (3.39) (3.70)
LIM -0.885 -9.055 ** 0.784
(0.15) (2.44) (0.11)
SEC -0.0010 -0.0006 -0.0007
(1.07) (0.64) (0.71)
IDE 0.0028 0.0081 -0.0014
(0.50) (1.17) (0.27)
NUM -0.0019 ** -0.0019 ** -0.0027 ***
(1.96) (1.97) (2.69)
HV -0.003 -0.002 0.0002
(0.76) (0.71) (0.07)
PCPI -0.008 *** -0.006 *** -0.010 ***
(3.30) (3.61) (3.35)
PRIME 0.266 *** 0.218 *** 0.292 ***
(6.07) (11.00) (5.08)
State Dummies Yes Yes Yes
Year Dummies Yes Yes Yes
N 8274 7919 6663
[R.sup.2] .194 .267 0.192
Absolute value of the t statistic in parentheses. A , **, and
** represent statistical significance at the 10, 5, and 1
percent level, respectively.
