Impact of state exemption laws on small business bankruptcy decision.

1. Introduction

Small businesses are a primary source of employment in the U.S. economy, employing over half of the private sector workforce. Small businesses are responsible for about two-thirds of all net new jobs created. According to the 2002 Small Business Economic Indicators (SBEI), (1) over 99.7% of the 5.7 million firms in the United States are classified as small- to medium-sized businesses. Hence, small businesses are a substantial contributor to economic growth in the United States.

Small business owners enjoy more flexibility and freedom to capitalize on profitable opportunities than their larger business competitors. Hence, they challenge the larger firms to be more efficient, which is ultimately beneficial to consumers. Unfortunately, small- and medium-sized businesses have a very low survival rate (SBEI 2002). In 2002 (on the heels of the 2001 recession), there were 550,100 small business births and 584,500 small business terminations. Both academics and public policy analysts have become increasingly concerned about the success of small businesses.

Recently there has been a surge in small business research, particularly surrounding small business credit supply and demand. Not surprisingly, much of the literature has been focused on changes in lending technologies (relationship (2) versus scored (3)), information processing technologies (soft versus hard), (4) loan size considerations, (5) and loan type considerations (6) as they relate to small business supply and demand. However, very few have actually examined why small- and medium-sized businesses have very low survival rates. Some studies have analyzed the sensitivities of Small Business Administration (SBA) guaranteed loans to macroeconomic changes industry risks, maturity structures (3, 7, and 15 year maturities), as well as lender and debt characteristics (e.g., Glennon and Nigro 2001, 2002; Dunsky and Pennington-Cross 2003). Meanwhile, Agarwal, Chomsisengphet, and Liu (2003) empirically assessed the significant importance of owner and firm characteristics in determining the risk of small business default while controlling for the macroeconomic and industry risks.

With bankruptcy filings continuing to rise significantly in recent years, many policy makers are turning their attention to bankruptcy laws. Specifically, Congress is considering reforming personal bankruptcy laws that, if passed, will significantly impact the demand for and supply of both consumer credit and small business funding. White (2001) concludes that this bankruptcy reform could potentially reduce small business ownership but increase the supply of small business credit and, in turn, affect the growth of the U.S. economy.

Small business owners have an incentive to file for personal bankruptcy when their indebtedness exceeds the value of their assets because both their personal and business debts can be discharged. (7) Though the bankruptcy exemption law is primarily designed for consumers, the personal bankruptcy exemption law is a de facto bankruptcy procedure for small business owners because the debt of a noncorporate firm is the personal liability of the entrepreneur/owner (Fan and White 2003). Hence, investigating the potential impact of the exemption law on small business bankruptcy decisions may provide some insight into how the bankruptcy laws should be reformed.

Further, Fan and White (2003) argue that while the expected return to creditors should be lower in states with higher exemption level upon small business shutdown, it is not entirely clear whether lenders will actually shut down a financially troubled small business. While they empirically find that small businesses in high-exemption states are more likely to be shut down, this positive relationship is statistically insignificant. The authors conclude that "additional research will be needed to determine if a significant relationship exists" (p. 563).

With micro-level data, we want to test whether small business borrowers act "strategically" in their bankruptcy decision in order to take advantage of the state bankruptcy laws. Our data are a unique panel data set of over 43,000 small business credit card holders over a two-year period (May 2000-May 2002). These panel data include information on small business bankruptcy filings as well as information on the business owners' financial and credit risk standing. These accounts are small business lines of credit with the following characteristics: the lines are under $100,000, contain a personal guarantee by the owner or principal, and were originated using scored lending technology that evaluated the creditworthiness of the business owner and not necessarily of the business. (8) Because small business credit card lending is secured only by personal assets of the owner, this part of the credit market ought to be theoretically most affected by exemption provisions, and any impact on small business' bankruptcy filing decisions as affected by the exemption laws should be quite noticeable.

To measure the impact of exemption laws on the likelihood of bankruptcy filings of small businesses using the loan-level data briefly described above, we estimate a proportional hazard model. Our empirical results suggest that an increase of $10,000 in the homestead exemptions will increase the likelihood of small business owners declaring bankruptcy by 8%. Moreover, our results also indicate there is a 4% rise in the risk of small business bankruptcy with a $1000 increase in personal property exemption levels.

The structure of the rest of the article is as follows: Section 2 provides an overview of the literature; section 3 describes the data and model specification; section 4 presents empirical findings; and section 5 offers concluding remarks.

2. Literature Review

White (2003) provides a theoretical and empirical literature review about the U.S. personal bankruptcy law. There are two personal bankruptcy procedures available to individuals in the United States. Debtors are allowed to choose between them. Under a Chapter 7 filing, unsecured debts such as credit card debt, installment loans, medical bills, and damage claims are discharged. Owners are not obliged to use any of their future earnings to repay their debt, but they must turn over all their assets above a certain level of state exemption (homestead and property) to the bankruptcy trustee. Because exemption varies widely across states, debtors residing in states with relatively higher exemptions will be able to retain more of their assets. Under a Chapter 13 filing, debtors do not have to give up any assets, but they must offer a plan to repay a portion of their debt with future income, usually over three to five years. Hynes (1998) describes Chapter 13 as the consumer analog of Chapter 11 reorganization. Whereas bankruptcy filers have a choice between Chapters 7 and 13, they have a financial incentive to choose Chapter 7 whenever their assets are less than their state's exemption, and thus can avoid repaying their debts completely (also see Fan and White 2003).

Homestead and personal property exemptions provide debtors in a bankruptcy filing with relief from creditors. Thus, discharging of debts provides debtors with a chance for a "fresh start." Homestead exemptions vary widely: from zero in two states to unlimited in seven states. About one-third of the states also allow their residents to choose between federal bankruptcy exemptions and state exemptions. (9)

Personal property exemptions also vary widely. For instance, Texas has the most generous personal property exemption level of $30,000, whereas Hawaii only allows $2000. Many states also allow married couples that file for bankruptcy to take higher exemptions, usually double.

We now review the literature on the impact of personal bankruptcy exemptions on consumer credit as well as small business credit.

Bankruptcy Exemptions and Household Credit

According to White (2003), "Bankruptcy is an important aspect of consumer credit markets, because whether consumers repay or default on their loans depends on whether the legal system punishes debtors who default and, if so, how severely." (p. 1) In this respect, Lin and White (2001) and White (2003) argue that the U.S. bankruptcy system is especially favorable toward the debtor. In fact, we do observe that the number of personal bankruptcy filings has doubled within a short span of seven years (1996-2002) from 700,000 to more than 1,400,000 (see Figure 1).

[FIGURE 1 OMITTED]

With bankruptcy filings rising significantly in the recent years, researchers are turning their attention to the potential impact of bankruptcy exemption laws on consumer credit markets. A number of empirical studies have focused on the supply of, as well as the demand for, credit. In particular, whether differences in states' bankruptcy exemption levels affect aggregate household credit (both secured and unsecured) has been studied by Gropp, Scholz, and White (1997); Berkowitz and Hynes (1999); Lin and White (2001); Fay, Hurst, and White (2002); Agarwal, Liu, and Mielnicki (2003); Berkowitz and White (2003); Chomsisengphet and Elul (2004); and White (2003).

Gropp, Scholz, and White (1997) find empirical support for their theoretical argument that in states with high rather than low bankruptcy exemptions, there is more credit rationing because debtors are more likely to default and file for bankruptcy. Berkowitz and Hynes (1999) argue that because of a positive wealth effect, a financially distressed homeowner filing for bankruptcy will be able to retain more of his assets in high-exemption states, which would enable him to continue paying his mortgage. Hence, we should expect that in states with generous exemptions, there should be lower risk of mortgage default and increased access to mortgage credit. On the other hand, Lin and White (2001) develop a theoretical model to show that in states with relatively higher exemptions, there should be more credit rationing in the mortgage market because lenders incur additional costs of default when a foreclosure occurs in conjunction with a bankruptcy filing. Berkowitz and Hynes and Lin and White test their predictions using a Home Mortgage Disclosure Act (HMDA) data set from the mid-1990s. Interestingly, both find the data to support their respective hypotheses.

Chomsisengphet and Elul (2004) argue that there is another channel through which exemptions can affect the mortgage market--the credit bureau scores. The authors construct a model to show that by ignoring the impact of exemptions on credit scores, lenders will tend to overstate the riskiness of borrowers from high-exemption states. They also empirically show that a regression that omits the credit score would indeed find that applicants from high-exemption states are more likely to be denied a mortgage; however, once they control for credit scores, exemptions are no longer relevant.

In other related studies, Agarwal, Liu, and Mielnicki (2003) empirically test whether homestead, personal property, and garnishment exemptions across states play an important part in consumers' decisions to become delinquent and eventually declare bankruptcy on their credit card debt. Their findings indicate that homestead exemption levels are statistically significant in determining an individual's decision to declare bankruptcy. Fay, Hurst, and White (2002) study the household bankruptcy decision using the personal survey income dynamics (PSID) data, which include information on bankruptcy filings. They conclude that households act "strategically" and declare bankruptcy only when the financial benefits outweigh the costs. Furthermore, they find little support for the prediction that households file for bankruptcy as a result of adverse events (unemployment, divorce, etc.) occurring, which in turn reduces households' ability to pay off their loans, (10) Interestingly, they also include a dummy variable for households owning a business in their study. They argue that self-employed individuals are less risk averse and consequently have a higher risk for filings. However, their results do not confirm this proposition. Though their paper empirically tests the effect of exemption law on the household bankruptcy decision while controlling for self-employed households, they do not explicitly test the effect of exemption law on the small business bankruptcy decision.

Bankruptcy Exemptions and Small Business Credit

Berkowitz and White (2003) argue that personal bankruptcy exemption laws not only apply to individuals but also to small businesses, irrespective of whether they are incorporated or unincorporated. Debts of an unincorporated firm are personal liabilities to the firm's owner; thus, the firm's debt at bankruptcy is also the owner's debt. On the other hand, an incorporated firm's limited liability implies that the firm's debt is legally distinct from the owner's debt. However, the authors argue that for small firms, lenders generally require that a firm's debt be guaranteed by its owner, and thus the lender may not view the incorporated/unincorporated distinction of small businesses as meaningful.

Furthermore, Fan and White (2003) develop an analytical model to show the impact of variation in state exemption levels on creditors' expected return. In particular, the authors show that creditors' expected return is lower in states with relatively higher exemptions because failing firms can protect their assets from creditors by shifting from nonexempt business categories to exempt personal categories. As a result, the authors predict that creditors are more likely to immediately shut down failing firms that are located in states with a high exemption level. However, the authors do not find statistical support for their hypothesis and conclude that additional research is needed.

3. Data and Methodology

Data

The data employed in this study are proprietary in nature. The accounts originated at several financial institutions that issue small business credit cards nationally, but they are managed at a single institution. The panel data set covers monthly small business information for around 63,014 accounts originated from May 2000 to May 2002. (11) Most of the businesses are very small, single-family owned, with no formal financial records. Furthermore, most of these small businesses predominantly secure their credit from debt financing as opposed to equity financing (also see Berger and Udell 1995). The data set has all the information that was collected at the time of account origination.

In addition, we also have access to all the performance information since May 2000. Hence, if an account originated in May 2000 and the account is still in good standing, then we would have performance data for a 24-month period. For each account, we have the borrower's quarterly bureau credit score updates, internal monthly behavior score updates, self-reported personal income at origination, credit limit, default indicators, bankruptcy indicator, and other information that characterizes an account's standing with the creditor. If the account closes and the account holder declares formal bankruptcy at any time, then a formal bankruptcy flag is created. Accounts with a flag indicating lost, stolen, never active, and closed because of fraud/death status are excluded from the analysis.

We include county unemployment rates as a macroeconomic control. Equally important, we also include state homestead and personal property exemption levels and garnishment levels (see Agarwal, Liu, and Mielnicki 2003). For the purpose of this analysis, we have excluded state-level social variables such as divorce, income, and health insurance coverage because of partial unavailability of the data for the time period covered. (12)

Because families who rent rather than own a house cannot use the homestead exemption levels, for the purpose of this study we only concentrate on small business owners who own a house. However, Fan and White (2003) argue correctly that households can purchase a house at any time before they decide to file for bankruptcy. (13)

Given all the above criteria, the number of accounts is reduced to about 43,324, with 404 bankruptcies. The number of bankruptcies is less than 1% of all outstanding loans. It is entirely possible that some accounts have defaulted on the loan (180 clays past due) but are in the process of or have declared bankruptcy; however, due to the short time frame of our study and the lag in reporting bankruptcy information to the credit bureau and a further lag in updating bureau information at the financial institution, the 404 bankruptcies are deemed to be underestimated. (14) However, to account for the possibility that some small businesses are either in the process of or have declared bankruptcy, we also estimate a separate model of small business default. (15)

Our sample of small business loans and bankruptcy filings is relatively well represented across the United States. Table 1 reports some of the descriptive statistics. Our small business sample has an average Fair Isaac & Co. (FICO) score of 715, with a standard deviation of 55: a lower FICO score indicates higher credit risk. The average credit line and the variable interest rates are $21,000 and 13.4%, respectively. On average, a small business in our data set has been in operation for 4.4 years. Finally, we also know that the borrowers in our data set have an average home mortgage debt of $102,684 and unsecured personal debt of $12,767.

The data set employed in our study is uniquely rich in comparison to the survey or account origination data employed in past research. As discussed in Gross and Souleles (2002), observing loan-level accounts over time has several advantages: (i) There is a greater likelihood of having a critical number of observations of even low-probability events like bankruptcy: (it) the time series nature of the data helps model bankruptcy in a dynamic as opposed to static model; and (iii) in contrast to data based on surveys of small businesses, measurement errors are less of a problem.

Methodology

We estimate a duration model for small business bankruptcy to investigate the impact of bankruptcy exemption levels on the decision of small business owners to file for bankruptcy. (16) Let [B.sub.i,t] indicate whether an account i becomes bankrupt in month t. Using the proportional hazard model, we estimate the conditional probability of bankruptcy at time t, assuming the consumer is current from inception up to time t - 1 (see, e.g., Kiefer 1988). For instance, the account could declare bankruptcy in the 18th month, then [B.sub.i,t] = 0 for months t = 1,..., 17, and [B.sub.i.18] = 1, and the rest of the observations will drop out of the sample. (17) The main specification would be as follows:

(1) [B.sub.i,t] = [h.sub.0](t)exp[[beta]'[Z.sub.i](t)],

where [h.sub.0](t) is the baseline hazard function at time t (the hazard function for the mean individual ith sample); we use calendar quarter dummies ([Time.sub.t]) of the account as a proxy for this baseline. [Z.sub.i](t) is a vector of time-varying and time-constant covariates; [beta] is the vector of unknown regression parameters to be estimated; and exp [[beta]'[Z.sub.i](t)] is the exponential distribution specification that allows us to interpret the coefficients on the vector of Z as the proportional effect of each of the exogenous variables on the conditional probability of "completing the spell," for example, small business filing for bankruptcy.

The time-varying as well as time-constant exogenous variables (known as covariates) that affect a small business owner's decision to file for bankruptcy can be specified as follows:

(2) [beta]'[Z.sub.i,t] = [[beta].sub.1][U.sub.i,t-6] + [[beta].sub.2]SI[C.sub.i] + [[beta].sub.3][OwnerAge.sub.i] + [[beta].sub.4]YI[B.sub.i] + [[beta].sub.5][X.sub.i,t-6] + [[beta].sub.6][L.sub.i] + [[epsilon].sub.i,t],

where [U.sub.i,t-6] denotes county unemployment rates lagged six months. Here, we use county unemployment rates to control for state-specific business-cycle effects. Because we are interested in the unemployment rate at the time the account was in good standing and not the time of actual bankruptcy, lagging this variable by six months will avoid any potential endogeneity of unemployment on bankruptcy. The two-digit SI[C.sub.i] represents dummies for various industry classifications and are used to control for any industry-specific effects. [OwnerAge.sub.i] designates the age of the small business owner at account origination. YI[B.sub.i] denotes the years in business dummy at the time of origination. (18) We include three variables of years in business, months on file, quarterly time dummies, and the age of the business owner in the equation to control for cohort, seasoning, time, and demographic differences across small business owners. As discussed in Degryse and Van Cayseele (2000), we use years in business as a control for time to buildup of public information and months on file as a control for time to buildup of private information. [X.sub.i,t-6] represents account-specific measures of risk. They include account balance, credit limit, interest rate, internal and external scores, and others. These variables are all lagged by six months to eliminate any potential endogeneity with small business bankruptcy (also see Gross and Souleles 2002). Finally, [L.sub.i] represents state-specific legal variables, including judicial foreclosure; statutory right of redemption; and homestead, property, and garnishment exemption levels.

4. Empirical Results

Exemption Laws and Small Business Bankruptcy Decision

We estimate several combinations and/or lag structures of the above general specification. Specifically, we estimate the model with various lag structures for the credit risk variables: (i) 6 lags, (it) 12 lags, and (iii) at origination. This helps alleviate any concern that the creditor, on the basis of small business performance, could endogenously change credit risk variables. We also conduct other robustness checks, such as augmenting the data to include a discrete set of dummy variables for garnishment, homestead, and personal property exemptions, as discussed in Fan and White (2003). Finally, we also include quadratic terms for the legal (homestead and property exemptions) and credit risk variables (internal and external scores). These variables change the results quantitatively but not qualitatively. In this article, we present results for only two specifications.

Table 2 presents empirical results on the impact of state exemption levels on the small business bankruptcy decision under two specifications. The main difference between the two models is that the first model assumes that homestead exemptions are continuous, while the second model assumes that they are discrete (as discussed in Fan and White 2003).

Let us first discuss the impact of the credit risk variables (e.g., credit scores, credit line, and annual percentage rate [APR]) on bankruptcy filings. As expected, the negative coefficients of both internal (payment behavior) and external (FICO) scores suggest that a decrease in internal or external scores significantly increases the likelihood of a small business owner filing for bankruptcy. The relationship between size of the credit line (Credit Line) and small business bankruptcy is also negative, but it is statistically insignificant. The coefficient for APR is positive and highly significant, implying that small business cards with higher APR have a higher probability of declaring bankruptcy. Some examples of the economic magnitudes of the predictive variables are as follows: A 10% drop in the external (FICO) score raises the probability of bankruptcy by 14%, while a 1% rise in the APR raises the probability of bankruptcy by 6%.

As for the macroeconomic effects, the county unemployment rate coefficient is positive but statistically insignificant. We also control for two-digit Standard Industrial Classification (SIC) by using dummy variables: although the results are not reported in the article, they isolate any industry-specific effects. Overall, our results suggest that the propensity to file for bankruptcy significantly varies on average by industry. Industries such as special trade contractors (SIC = 17), electronic and other electric equipment (SIC = 36), general merchandise stores (SIC = 53), and eating and drinking places (SIC = 58) have a higher propensity to declare bankruptcy. On the other hand, industries such as furniture and home furnishing stores (SIC = 57), health services (SIC = 80), and social services (SIC = 83) have a lower propensity to declare bankruptcy. The coefficients for the SIC codes are jointly significant, implying the considerable presence of industry-specific effects.

Next, we control both for calendar time, years in business, and account age. It is worth noting that age of an account could be different than years in business. For instance, an account that has been two years in business could originate at the financial institution on the same day as a different account that has been three years in business. For a firm with one extra year in business, the margin impact of bankruptcy filing falls by 6%. Since one would expect that younger business owners might be less risk averse, we also control for the business owner's age. While we do not report the coefficient values for these variables, they are jointly significant for each variable and directionally intuitive. For example, owners of small businesses that have been in business longer are less likely to file for bankruptcy. As for one of the main variables of interest, property exemptions are positive and statistically significant for both models 1 and 2. Although Fan and White (2003) do not consider the property exemption levels as part of their study, our results indicate that small business entrepreneurs from states with high property exemptions are more likely to file for bankruptcy.

As for the state homestead exemption levels, we create a continuous variable for the homestead exemption levels in model 1, and we follow Fan and White (2003) and construct quartile dummy variables for homestead exemptions to represent each quartile of the distribution except the lowest in model 2. Furthermore, we also construct a separate dummy for the unlimited homestead exemptions for model 2. In comparison with the results of Fan and White, our results in model 2 indicate that the homestead exemption dummies for the second and third quartiles are negative but statistically insignificant. On the other hand, the dummies for the fourth quartile and the unlimited exemption states are positive and statistically significant, indicating that small businesses from states with very generous homestead exemptions are more likely to file for bankruptcy. These results imply that small business owners residing in the relatively more generous homestead exemption states are more likely to act "strategically" in declaring bankruptcy. (19)

It is also worth noting that despite differences in methodology (hazard vs. random effects probit), time period, frequency (1993:I-1995:IV versus 2000:6-2002:6), and data type (survey versus account level), the results of our study and that of Fan and White (2003) are fairly consistent in that both studies find a positive relationship between homestead exemptions and small business bankruptcy. However, while this relationship is insignificant in their study, it is statistically significant in our study. Specifically, our empirical results suggest that an increase of $10,000 in the homestead exemptions will increase the likelihood of small business owners declaring bankruptcy by 8%. Moreover, our results also indicate that the probability of small business owners filing for bankruptcy rises by 4% with a $1000 increase in personal property exemption levels. These results are estimated using the baseline survival function at the mean values for all other variables and predict the rise in the probability of bankruptcy over a one-year time horizon.

Furthermore, three other legal variables measuring the effect of state foreclosure laws also carry the expected signs. States with statutory right of redemption have higher risks of bankruptcy; however, the coefficient is statistically insignificant. States that require judicial foreclosure prolong the foreclosure process and thereby should have lower risks of bankruptcy; however, the estimated coefficient is insignificant. Finally, states that do not allow deficiency judgment should have a higher risk of small business filing for bankruptcy, since this prohibits the lender from collecting unpaid balances on personal assets. (20) Clauretie and Herzog (1990) and Pence (2003) also report similar results. We also control for garnishment allowance in case of bankruptcy. The results indicate that states with high garnishment allowance have lower probability of bankruptcy, although the coefficient value is statistically insignificant.

We also include dummy variables for both total secured debt (mortgage, auto, and home equity loans/lines) and total unsecured debt (unsecured lines of credit) of the small business owner as reported at the credit bureau and updated quarterly. We construct five categories for each type of debt. For unsecured debts, the five categories are (i) $0 to $1000: (ii) $1000 to $5000: (iii) $5000 to $10,000; and (iv) greater than $10,000. For secured debts, the categories are (i) $0 to $10,000; (ii) $10,000 to $25,000; (iii) $25,000 to $50,000: and (iv) greater than $50,000. Although we tried several other cutoff amounts, including continuous variables, for both secured and unsecured debt, the results did not change substantially. Our results show that small business owners with higher unsecured (secured) debt are more (less) likely to declare bankruptcy; the coefficients for these variables are jointly significant.

Exemption Laws and Small Business Investment Decision

An increase in bankruptcy exemptions not only provides small business owners with wealth protection when their businesses fail, but may also encourage small business owners to take higher risks in their business decisions and induce lenders to charge higher interest rates. It will be interesting to know (i) whether those small business owners living in states with higher exemptions indeed take more risky investment projects and (ii) whether lenders charge higher APR or offer lower credit lines at account origination for those small business owners who live in higher exemptions states. This section provides some insights to these important questions.

Owing to the lack of adequate information about the choice of a small business owner's investment project decision, we use the volatility of an account's payment during the study period as a proxy for the riskiness of the project. As documented within the small business literature (e.g., Strahan 1999; Agarwal, Chomsisengphet, and Driscoll 2004), small businesses demand lines of credit to meet recurring expenses as opposed to financing long-term investments. Hence, we would expect a low-risk business owner to draw up and down the line fairly uniformly. However, a high-risk business might experience higher payment volatility. Hence, we use the following regression model to determine whether exemption levels affect the riskiness of small business projects:

(3) [V.sub.i] = [[beta].sub.1][X.sub.i] + [[beta].sub.2]SI[C.sub.i] + [[beta].sub.3] YI[B.sub.i] + [[beta].sub.4] [Owner[Age.sub.i] + [[beta].sub.5] [L.sub.i] + [[epsilon].sub.i,

where [V.sub.i] is the standard deviation of monthly payment as ratio of total outstanding debt over a two-year period for a small business i. (21) [X.sub.i] are the determinants of customer credit risk at account origination (external credit score, bureau debt, bureau credit line); SI[C.sub.i] represents dummies for various industry classifications. YI[B.sub.i], denotes years in business. Owner[Age.sub.i] designates the age of the small business owner at account origination. Finally, [L.sub.i] represents state-specific legal variables like homestead and property exemptions.

Table 3 reports the results. SIC and OwnerAge dummies are not reported. Some of the SIC dummies are statistically significant and indicate that higher risk industries are positively correlated with payment volatility. OwnerAge dummies are statistically insignificant. Both APR and external score are statistically significant and positive--borrowers with a higher APR and higher credit quality have higher payment volatility. Finally, homestead and property exemptions are statistically insignificant, indicating that business owners located in states with relatively higher exemptions do not invest in riskier investment projects.

Impact of Exemption Laws on APR and Credit Lines

Finally, we conduct additional analysis to determine whether lenders will charge higher interest rates or offer lower credit lines at account origination to small business owners who live in higher exemption states. Specifically, we estimate the following linear regression model to determine the impact of exemption levels on interest rates and lines of credit at account origination:

(4) [Y.sub.i] = [[beta].sub.1][X.sub.i] + [[beta].sub.2] SI[C.sub.i] + [[beta].sub.3] YI[B.sub.i] + [[beta].sub.4] [L.sub.i] + [[epsilon].sub.i],

where [Y.sub.i] can either be APR or Line Amount at account origination of a small business i. Definitions of other variables ([X.sub.i], SI[C.sub.i], YI[B.sub.i], [L.sub.i]) in the regression are the same as in Equation 2.

Table 4 presents the results for two different models that test the impact of bankruptcy exemption laws on APR and line amount at account origination. The results show that a higher credit score leads to lower APR and higher line amount. A low bureau debt and high bureau credit line (indications of high creditworthiness of the business owner) lead to lower APR and higher line amount. Finally, the coefficient for years in business also has statistically significant impact on APR and line amount. The longer the owner has been in business, the lower the APR is and the higher the line amount is. (22) These results are consistent with the risk-based pricing hypothesis proposed in Edelberg (2003).

On the other hand, both homestead and property exemptions are statistically insignificant and do not help predict either APR or line amount, implying that the lender may not endogenously determine the APR or line amount based on applicants' residence. These results are consistent with Berkowitz and White (2003), who also conclude that variations in bankruptcy exemption levels do not cause loan size to be smaller or interest rates to rise. There are several possible explanations for these results. The first proposed by Berkowitz and White (2003) contends that financial institutions may practice an extreme form of credit rationing, that is, completely decline credit. Another possible explanation could be that credit rationing along state lines could be considered "redlining" and deemed illegal.

Finally, there is concern that credit scores may be subject to the effect of exemption laws and, in turn, may be the reason that the coefficients for the exemption laws are statistically insignificant. To address this concern, we estimate the correlation between scores and exemptions. The correlation coefficient of -0.13 (p value = 0.0403) is marginally significant at the conventional 5% level, suggesting that on average higher exemption states have lower credit scores. These results are consistent with Chomsisengphet and Elul (2004), who argue that credit scores for consumers are lower, on average, in states with higher exemptions.

5. Conclusion

Over the past five years, there has been tremendous interest in the area of small business credit and default as a result of a dramatic rise in personal bankruptcy filings. Advocates of the "strategic" consumer behavior attempting to explain the rise in bankruptcy filings argue that the surge in bankruptcy filings is in part due to higher homestead and personal property exemption levels that enable those in bankruptcy filings to retain more of their assets. The purpose of this study is to provide an answer to the question: Do state exemption laws affect the decision of small business owners to file for bankruptcy?

Using panel data for over 43,000 small business credit card lines of credit that originated using scored lending technology under $100,000, we estimate a proportional hazard model to assess the effect of state exemption laws on small business bankruptcy decision. Our empirical results suggest that a rise of $10,000 in the homestead exemptions will increase the probability of small business owners who are declaring bankruptcy by 8%. Moreover, our results also indicate that the likelihood of small business owners filing for bankruptcy increases by 4% given a $1000 increase in personal property exemption levels.

Our study--although the first to use loan-level panel data to empirically measure the impact of state exemption laws on small business bankruptcy decision--has several limitations. The results only look at the small business credit cards for scored customers under $100,000 lines of credit from a single financial institution. These results could vary depending on institutional differences, lending technology, and loan size, as well as loan type differences (see, e.g., Agarwal, Chomsisengphet, and Liu 2003). Despite the above limitations, we think that these results make a significant contribution to the debate surrounding exemption laws and small business bankruptcy. Our results suggest that the small business owners might act "strategically," to take advantage of exemption laws in their decision to file for bankruptcy.

Appendix

Proportional Hazard Model Measuring the Response of State Exemptions
Laws on Small Business Default (a)

                                               Model 1

                                 Coefficient  Standard
Independent Variables               Value       Error     p Value

External [score.sub.t-6]          -0.0091500    0.00098   <0.0001
Internal [score.sub.t-6]          -0.0024300    0.00021    0.0001
Line [amount.sub.t-6]             -0.0000070    0.00001    0.5787
AP[R.sub.t-6]                      0.2972100    0.02529    0.0001
Homestead 5K-10K
Homestead 10K-25K
Homestead >25K
Homestead 1 million
Homestead                          0.0000005   0.0000002   0.0251
Property                          -0.0000343   0.000014    0.0109
Deficiency judgment                0.4087100   0.189930    0.0314
Garnishment                       -0.0051100   0.009320    0.5839
Statutory right of redemption     -0.0157500   0.218110    0.9424
Judicial foreclosure              -0.2211100   0.267110    0.4078
Unsecured debt 1K-5K               0.0144200   0.214750    0.9464
Unsecured debt 5K-10K              0.0619200   0.235420    0.7925
Unsecured debt >10K                0.4983700   0.174650    0.0043
Secured debt 10K-25K              -0.0046100   0.189320    0.9806
Secured debt 25K-50K              -0.2238300   0.237300    0.3456
Secured debt >50K                 -0.3293700   0.138600    0.0175
County [unemployment.sub.t-6]      0.0280400   0.023060    0.2430
Two-digit SIC dummies                Yes
Years in business dummies            Yes
Small business owner age
  dummies                            Yes
Quarter dummy                        Yes
Log likelihood/pseudo [R.sup.2]     2231        0.35
Number of
  observations/defaults            219,993       894

                                               Model 2

                                 Coefficient  Standard
Independent Variables               Value       Error    p Value

External [score.sub.t-6]           -0.009190    0.00099   0.0001
Internal [score.sub.t-6]           -0.002420    0.00021   0.0001
Line [amount.sub.t-6]              -0.000007    0.00001   0.5646
AP[R.sub.t-6]                       0.296250    0.02533   0.0001
Homestead 5K-10K                   -0.159760    0.05442   0.0074
Homestead 10K-25K                  -0.099910    0.29381   0.7338
Homestead >25K                      0.066520    0.03649   0.0802
Homestead 1 million                 0.428050    0.16353   0.0034
Homestead
Property                           -0.000037    0.00001   0.0095
Deficiency judgment                 0.443610    0.21513   0.0392
Garnishment                        -0.010070    0.01113   0.3658
Statutory right of redemption       0.018760    0.23694   0.9369
Judicial foreclosure               -0.232860    0.28780   0.4185
Unsecured debt 1K-5K                0.013890    0.21502   0.9485
Unsecured debt 5K-10K               0.059480    0.23553   0.8006
Unsecured debt >10K                 0.493480    0.17469   0.0047
Secured debt 10K-25K               -0.007560    0.19004   0.9683
Secured debt 25K-50K               -0.235290    0.23794   0.3227
Secured debt >50K                  -0.329210    0.13861   0.0175
County [unemployment.sub.t-6]       0.028270    0.02410   0.2493
Two-digit SIC dummies                Yes
Years in business dummies            Yes
Small business owner age
  dummies                            Yes
Quarter dummy                        Yes
Log likelihood/pseudo [R.sup.2]     2319        0.37
Number of
  observations/defaults            219,993       894

(a) The table reports results of a hazard model of small business
default using monthly account level data from May 2000 to May 2002.
Explanatory variables include account-specific risk factors-external
score, internal score, APR, line amount, age (owner and firm), total
unsecured and secured debt; legal factors-deficiency judgment,
garnishment, statutory right of redemption, judicial foreclosure
requirement, property and homestead exemptions; macroeconomic risk
factors-county unemployment; industry risk factors-SIC dummies.
Subscript t - 6 represents the control variables six months prior
to default.

Table 1. Descriptive Statistics (a)

                             Mean      Standard
                                       Deviation

External (FICO) score          715          55
Internal (payment
 behavior) score               723          78
Credit line amount ($)      21,000       6,058
Total unsecured debt ($)    12,767      17,760
Total mortgage debt ($)    102,685     160,800
APR (%)                         13.40        5.36
Years in business                4.40        4.70
County unemployment              5.10        2.14
 rate (%)

This table provides summary statistics for some of
the key variables. Higher scores (external and internal)
imply lower probability of default. Unsecured and secured
debt measures total debt as reported by the credit bureau.

Table 2. Proportional Hazard Model Measuring the Response of State
Exemptions Laws on Small Business Bankruptcy (a)

                                                Model 1

                                 Coefficient   Standard
Independent Variables               Value        Error     p Value

External [score.sub.t-6]          -0.0091700    0.0009893  0.0001
Internal [score.sub.t-6]          -0.0024200    0.0002159  0.0001
Line [amount.sub.t-6]             -0.0000077    0.0000126  0.5415
AP[R.sub.t-6]                      0.2952300    0.0251500  0.0001
Homestead 5K-10K
Homestead 10K-25K
Homestead >25K
Homestead 1 million
Homestead                          0.0000005    0.0000002  0.0293
Property                           0.0000392    0.0000126  0.0018
Deficiency judgment                0.3252700    0.1712000  0.0574
Garnishment                       -0.0085800    0.0085800  0.3173
Statutory right of
  redemption                      -0.0139100    0.2173300  0.949
Judicial foreclosure
  required                         0.0004191    0.1537200  0.9978
Unsecured debt 1K-5K               0.0167900    0.2147600  0.9377
Unsecured debt 5K-10K              0.0619600    0.2354300  0.7924
Unsecured debt >10K                0.4894300    0.1745700  0.0051
Mortgage debt 10K-25K             -0.0125800    0.1892300  0.947
Mortgage debt 25K-50K             -0.2253400    0.2373900  0.3425
Mortgage debt >50K                -0.3247500    0.1384800  0.019
County [unemployment.sub.t-6]      0.0279800    0.0269300  0.2988
Two-digit SIC dummies                Yes
Years in business
  dummies                            Yes
Small business
  owner age dummies                  Yes
Quarter dummy                        Yes
Log likelihood/pseudo [R.sup.2]     2123         0.34
Number of
  observations/bankruptcy          215,203        404

                                                Model 2

                                 Coefficient   Standard
Independent Variables               Value        Error     p Value

External [score.sub.t-6]          -0.0091900    0.0009928  0.0001
Internal [score.sub.t-6]          -0.0024100    0.0002165  0.0001
Line [amount.sub.t-6]             -0.0000082    0.0000126  0.5174
AP[R.sub.t-6]                      0.2952000    0.0252200  0.0001
Homestead 5K-10K                  -0.0096700    0.2455700  0.9686
Homestead 10K-25K                 -0.0235100    0.0190800  0.14304
Homestead >25K                     0.1870900    0.0702400  0.04361
Homestead 1 million                0.5069200    0.2023300  0.0245
Homestead
Property                           0.0000418    0.0000139  0.0026
Deficiency judgment                0.4039400    0.2117400  0.0564
Garnishment                       -0.0108900    0.0110000  0.3219
Statutory right of
  redemption                      -0.0205000    0.2336800  0.9301
Judicial foreclosure
  required                         0.0183900    0.1588500  0.9078
Unsecured debt 1K-5K               0.0193400    0.2149500  0.9283
Unsecured debt 5K-10K              0.0567200    0.2355300  0.8097
Unsecured debt >10K                0.4861700    0.1746900  0.0054
Mortgage debt 10K-25K             -0.0081300    0.1901500  0.9659
Mortgage debt 25K-50K             -0.2321600    0.2378700  0.3291
Mortgage debt >50K                -0.3244600    0.1384400  0.0191
County [unemployment.sub.t-6]      0.0280100    0.0270300  0.3001
Two-digit SIC dummies                Yes
Years in business
  dummies                            Yes
Small business
  owner age dummies                  Yes
Quarter dummy                        Yes
Log likelihood/pseudo [R.sup.2]     2139         0.36
Number of
  observations/bankruptcy          215,203        404

(a) The table reports results of a hazard model of small business
bankruptcy using monthly account level data from May 2000 to May 2002.
Explanatory variables include account-specific risk factors (external
score, internal score, APR, line amount, age [owner and firm], total
unsecured and secured debt); legal factors (deficiency judgment,
garnishment, statutory right of redemption, judicial foreclosure
requirement, property and homestead exemptions); macro risk factors
(county unemployment); and industry risk factors (SIC dummies).
Subscript t - 6 represents the control variables six months prior to
bankruptcy.

Table 3. Response of State Exemptions Laws on Business Payment
Volatility over a Two-Year Period (a)

                                                Standard
                            Coefficient Value    Error     t-Statistic

Intercept                      -2.8664400      3.3653260      -0.85
Credit line                    -0.0000118      0.0000246      -0.48
Total bureau debt               0.0000012      0.0000008       1.47
Years in business               1.2008140      0.8459816       1.42
Years in business (square)     -0.2255190      0.4457271      -0.51
Property exemptions             0.0000273      0.0000222       1.22
Homestead exemptions            0.0000001      0.0000001       0.25
External score                  0.0067415      0.0028680       2.35
APR                             0.0394364      0.0178757       2.21
SIC dummies                        Yes
Age dummies                        Yes
Adjusted [R.sup.2]                0.07
Number of observations           43,324

(a) The table reports results of an ordinary least square regression
of payment volatility at account origination. The key explanatory
variables include property and homestead exemption levels.

Table 4. Response of State Exemption Laws on APR and Line Amount
at Account Origination (a)

                                   Dependent Variable: APR

                            Coefficient   Standard
                               Value       Error     t-Statistic

Intercept                    16.709500   0.61166000     27.32
Bureau credit line          -0.00000005  0.00000011     -0.44
Bureau debt                  0.00000784  0.00000178      4.42
Years in business            0.61203000  0.23049000      2.66
Years in business (square)  -1.21081000  0.10982000    -11.03
Property exemptions         -0.00004308  0.00006200     -0.69
Homestead exemptions        -0.00000060  0.00000052     -1.15
External score              -0.01368000  0.00081600    -16.76
SIC dummies                     Yes
Adjusted [R.sup.2]             0.02
Number of observations        43,324

                              Dependent Variable: Line Amount

                            Coefficient   Standard
                               Value       Error     t-Statistic

Intercept                  -14551.000     463.043     -31.42
Bureau credit line              0.00037     0.00008     4.4
Bureau debt                    -0.02456     0.00136   -18.09
Years in business            1402.63177   176.462       7.95
Years in business (square)   2891.8799     83.154      34.78
Property exemptions             0.02279     0.047       0.48
Homestead exemptions            0.00002     0.00009     0.26
External score                 30.57888     0.618      49.48
SIC dummies                     Yes
Adjusted [R.sup.2]             0.14
Number of observations        43,324

(a) The table reports results of ordinary least square regressions of
APR and credit line amount at account origination. The key explanatory
variables include property and homestead exemption levels.

(1) See http://www.sba.gov/advo/stats/sbei02.pdf.

(2) Studies have looked at the relationship between banks and access to small business lending. Recent studies have looked at the existence of the relationship (Cole 1998), the breadth of the relationship (Mester, Nakamura, and Renault 2001), and the exclusivity of the relationship (Berger, Klapper, and Udell 2001: Ongena and Smith 2001).

(3) Studies have looked at the impact of scored lending of small business loans. Frame, Srinivasan, and Woosley (2001) and Berger, Frame, and Miller (2002) conclude that scored lending tends to increase the portfolio share of small business loans.

(4) See Petersen (2002) and Stein (2002).

(5) Berger, Frame, and Miller (2002) conjecture that credit availability, price, and risk for small business loans below and above 100K differ in many respects.

(6) Berger and Udell (1995) differentiate between small business loan commitments and lines of credit.

(7) It is important that we clarify the difference between default and bankruptcy. An account is declared in default by the lender based either on a violation of one of the loan covenants or sufficiently severe delinquency that the lender begins legal proceedings to recover its funds. Moreover, default does not necessarily lead to bankruptcy, although in most cases borrowers will choose to file for bankruptcy to forestall the lender's legal actions. In this study we exclusively focus on bankruptcy fliers, but, as indicated earlier, because most borrowers in default choose to file for bankruptcy, we will validate our empirical estimates on defaulted accounts as well check for robustness of the results.

(8) Although it might appear that these loans are originated as consumer credits, they are managed as small business loans.

(9) This is due to the 1978 Congressional attempt to remove the state variation of exemptions. However, a political compromise led to a bill that granted a debtor with a choice of either the new federal exemptions or state-level exemptions.

(10) See Sullivan. Warren, and Westbrook (1989).

(11) In some cases bankruptcy resolution can take a long time, so it is possible that a 24-month window may be insufficient. However, we still can find interesting results with the current data limitations.

(12) For completeness we did include some of the social variables and estimated the model on a shorter time period (May 2000-December 2001), and the results are qualitatively the same.

(13) As suggested by the referee, we estimate a separate model with the inclusion of the renters. We interact the renters and owners with the homestead exemption variable. As expected, the coefficients for the renters interacted with homestead exemption levels is statistically insignificant, whereas the coefficients for the owners is statistically significant.

(14) However, according to the SBEI (2002), while personal bankruptcies reached a new record in 2002, small business bankruptcies dropped by 2.1% from 2001. The SBEI argues that a drop in interest rates and a cut in labor force (removing inefficient operations) account for the drop in small business bankruptcies.

(15) In our data set, there are 894 accounts that defaulted. Some of them will invariably declare bankruptcy, others might enter into forbearance plans with the lender, while others will simply default without declaring formal bankruptcy. The data set does not lend itself to distinguish borrowers who enter the forbearance plans versus those who default without declaring bankruptcy. Finally, we do not know the percentage loan recovery either in the event of default or bankruptcy.

(16) Shumway (2001) identifies three reasons for using a duration model as opposed to a static model: (i) Static models fail to control for firm period risk, Oil duration models incorporate lime-varying covariates, and (iii) duration models produce more efficient out-of-sample bankruptcy estimates by using more data.

(17) This is also the most common form of incomplete data and is also defined as right censored. Accordingly, all customers start at time t = 0 and a customer could drop out of the sample because they close the account in good standing or even stay current till the end of two years without declaring bankruptcy.

(18) We construct dummies as opposed to a continuous variable for years in business because the financial institution used discrete cutoffs of years in business for portfolio management. We construct five dummy variables as follows. For a firm that has been in business less than I year, YIB_D1=1 else=0. For a firm that has been in business more than 1 year but less than 2.5 years, YIB_D2=1 else=0. For a firm that has been in business more than 2.5 years but less than 5 years, YIB_D3=1 else=0. For a firm that has been in business more than 5 years but less than 10 years, YIB_D4=1 else=0. Finally, for a firm that has been in business more than 10 years, YIB_D5=1 else=0. We also estimate the model with a continuous variable for YIB and the results are the same.

(19) We also estimate a model for small business defaults as opposed to bankruptcies. As discussed in Endnote 7, because of the possibility of lags in reporting bankruptcy to the lenders and default being a precursor to bankruptcy we should find similar results. As reported in the Appendix, the results are remarkably similar both qualitatively and quantitatively.

(20) We define the deficiency judgment variables as 1 if the state does not allow deficiency judgment.

(21) We also estimate the model with the dependent variable being the standard deviation of monthly purchases. The results are qualitatively the same.

(22) Since years in business captures the buildup of public information as discussed in Degryse and Van Cayseele (2000), it implies that the longer the owner is in business the lower his/her chance of default and bankruptcy.

References

Agarwal, S., S. Chomsisengphet, and J. Driscoll. 2004. Loan commitments and private firms. Unpublished paper. Finance and Economics Discussion Series 2004-27. Board of Governors of the Federal Reserve System.

Agarwal, S., S. Chomsisengphet, and C. Liu. 2003. Determinants of small business default: Owner versus firm characteristics. Unpublished paper. University of Nevada. Reno.

Agarwal, S., C. Liu, and L. Mielnicki. 2003. Exemption laws and consumer delinquency and bankruptcy behavior: An empirical analysis of credit card data. Quarterly Review of Economic and Finance 43:273-89.

Berger, A. N., W. S. Frame, and N. H. Miller. 2002. Credit scoring and the availability, price, and risk of small business credit. Unpublished paper, Federal Reserve Bank of Atlanta, Working Paper 2002-6.

Berger, A. N., L. F. Klapper, and G. F. Udell. 2001. The ability of banks to lend to informationally opaque small businesses. Journal of Banking and finance 25:2127-67.

Berger, A. N., and G. F. Udell. 1995. Relationship lending and lines of credit in small lima finance. Journal of Business 68:351-81.

Berkowitz, J., and R. Hynes. 1999. Bankruptcy' exemptions and the market for mortgage loans. Journal of Law and Economics 42:809-30.

Berkowitz, J., and M. J. White. 2004. Bankruptcy and small firms" access to credit. The RAND Journal of Economics 35:69-84.

Chomsisengphet, S., and R. Elul. 2004. Personal bankruptcy exemptions, credit history and the mortgage market. Unpublished paper, Office of the Comptroller of the Currency, Washington. DC.

Clauretie, T. M., and T. N. Herzog. 1990. The effect of state foreclosure laws on loan losses: Evidence from the mortgage insurance industry. Journal of Money, Credit and Banking 22:221-33.

Cole, R. A. 1998. The importance of relationships to the availability of credit. Journal of Banking and Finance 22:959-77.

Degryse, H., and P. Van Cayseele. 2000. Relationship lending within a bank-based system: Evidcnce from European small business data. Journal of Financial Intermediation 9:90-109.

Dunsky, R., and A. Pennington-Cross. 2003. Estimation. deployment, and backtesting of default and prepayment equations. Internal Working Paper, Office of Federal Housing Enterprise Oversight, Washington, DC.

Edelberg, W. 2003. Risk-based pricing of interest rates in household loan markets. Unpublished paper, Federal Reserve Board.

Fan, W., and M. J. White. 2003. Personal bankruptcy and the level of entrepreneurial activity. Journal of Law and Economics 46:543-67.

Fay. S., E. Hurst. and M. J. White. 2002. The household bankruptcy decision. American Economic Review 92:706-18.

Frame, W. S., A. Srinivasan, and L. Woosley. 2001. The effect of credit scoring on small business lending. Journal of Money Credit and Banking 33:813-25.

Glennon, D., and P. Nigro. 2001. An analysis of SBA small business loan default by maturity structure. Working Paper, Office of the Comptroller of the Currency, Washington, DC.

Glennon, D., and P. Nigro. 2002. Measuring the default risk of small business loans: A survival analysis approach. Working Paper, Office of the Comptroller of the Currency, Washington, DC.

Gropp, P. J., K. Scholz. and M. J. White. 1997. Personal bankruptcy and credit supply and demand. Quarterly Journal of Economics 112:217-51.

Gross, D. B., and N. S. Souleles. 2002. All empirical analysis of personal bankruptcy and delinquency. Review of Financial Studies 15:319-47.

Hynes, R. M. 1998. Three essays on exemptions and consumer bankruptcy. Ph.D. thesis. University of Pennsylvania, Philadelphia, PA.

Kiefer, N. M. 1988. Economic duration data and hazard functions. Journal of Economic Literature 26:646-79.

Lin, Y. E., and M. J. White. 2001. Bankruptcy and the market for mortgage and home improvement loans. Journal of Urban Economics 50:138-62.

Mester, L. L, L. I. Nakamura, and M. Renault. 2001. Checking accounts and bank monitoring. Unpublished paper, Wharton Financial Institutions Center.

Ongena, S., and D. Smith. 2001. Empirical evidence on the duration of banking relationships. Journal of Financial Economics 61:449-75.

Pence, K. 2003. Foreclosing on opportunity: State laws and mortgage credit. Unpublished paper, Federal Reserve Board.

Petersen, M. A. 2002. Information: Hard and soft. Unpublished paper, Northwestern University.

Shumway, T. 2001. Forecasting bankruptcy more accurately: A simple hazard model. Journal of Business 74:101-24.

Small Business Administration. 2002. Small business economic indicators for 2002. Accessed 10 January 2003. Available http:// www.sba.gov/advo/stats/sbei02.pdf.

Sullivan, T. A., E. Warren, and J. L. Westbrook. 1989. As we forgive our debtors. New York: Oxford University Press.

Stein, J. 2002. Information production and capital allocation: Decentralized versus hierarchical flints. Journal of Finance 57:1891-922.

Strahan, P. E. 1999. Borrower risk and the price and nonprice terms of bank loans. Unpublished paper, Federal Reserve Bank of New York.

White, M. J. 2001. Bankruptcy and small business. Regulation 18:18-20.

White, M. J. 2003. Bankruptcy and consumer credit in the U.S. Unpublished paper, University of California at San Diego.

Sumit Agarwal, * Souphala Chomsisengphet, ([dagger]) Chunlin Liu, ([double dagger]) and Lawrence Mielnicki [section]

* Bank of America, Mail Stop MD9-978-03-02. 1101 Wootton Parkway, Rockville, MD 20814, USA: E-mail [e-mail].

([dagger]) Risk Analysis Division, Office of the Comptroller of Currency, 250 E Street, SW, Washington, DC 20219. USA: E-mail [e-mail]: corresponding author.

([double dagger]) College of Business Administration, University of Nevada, Reno, Reno, NV 89557, USA: E-mail [e-mail].

([section]) De Lage Landen Financial Services. 1111 Old Eagle School Road, Wayne, PA 19087, USA: E-mail lrnielnicki@ leasedirect.com. The authors would like to thank Jim Papadonis and Margaret Burgess for their support of this research project. We would also like to thank Laura Razzolini (editor), Nick Souleles, and two anonymous referees for helpful comments. We are grateful to Diana Andrade, Ron Kwolek, and Greg Pownell for excellent research assistance. The views expressed in this research are those of the authors and do not represent the policies or positions of the Office of the Comptroller of Currency, of any offices, agencies. or instrumentalities of the United States Government, or of the Bank of America.

Received August 2003; accepted June 2004.