Financial Update (April-June 1999)


Cover Story

Advice to the Fed

Strategies for the New Millennium


Year 2000

Did You Know?

Data Bank

The Docket

Credit Scoring in Small Business
Lending: Bane or Blessing?

A s banks get bigger, some customers are finding that bank services are losing their personal touch. This trend may spell changes for small businesses, which have traditionally relied on an established relationship with a local bank as an avenue for obtaining credit.

Research on the characteristics of small business lending has pointed out the importance of the relationship between small business borrowers and lenders. Lenders face more uncertainty about the riskiness of small business loans because less information is publicly available about small businesses than about larger firms. To deal with this "information asymmetry" problem, banks may either ration small business loans — by charging higher rates or by limiting credit — because of the potentially higher risks involved or develop additional information from a relationship with the borrowers. For example, a bank that holds a small business applicant's deposits can evaluate the firm's cash flow by analyzing its account history.

Small businesses have traditionally relied on established relationships with local banks as an avenue for obtaining credit, but credit scoring may end the need for such relationships.

In low- and moderate-income (LMI) areas, small business lending faces additional hurdles. Banks may have concerns about making small business loans in LMI areas because they may have fewer branches that can establish business relationships in these areas. Banks may also be wary of LMI areas in general as places where businesses may have difficulty thriving or even surviving. On the other hand, the pressure exerted by the Community Reinvestment Act for banks to lend in LMI areas may offset the loan-deterring characteristics of these areas' businesses.

Credit Scoring Small Business Loans

One possible solution to some of the information asymmetry problems banks face in small business lending, particularly in LMI areas, may be the use of credit scoring. In the past few years, some banks have begun using this method to assess small business loans. Credit scoring, used for a decade or so in evaluating consumer loans and mortgages, is an automated method that analyzes a large sample of past borrowers in order to calculate the probability that a loan applicant with certain specific characteristics will default.

Access to Capital and Credit
Small business lending was the focus of a conference on business access to capital and credit, sponsored by the community affairs officers of the Federal Reserve System. The conference, held March 8–9, brought together economists, scholars, bankers and bank regulators to discuss papers and offer insight into topics surrounding small business lending.

One session at the conference focused on Community Reinvestment Act (CRA) data on small business lending. The papers in this session analyzed 1997 CRA data on lending to small businesses and farms, small business lending in intra-metropolitan areas, and access to capital issues. A second session at the conference examined access to credit for minority-owned businesses. Topics in this session included discrimination in the small business credit market; evidence of competition, credit rationing and discrimination on small business credit markets; and an examination of how race, geography, risk and market structure affect small business finance.

A two-part session on the small business lending relationship focused on borrowing from large and small banks and the effects of bank consolidation on access to lending. Papers presented in this session examined small business finance in two Chicago minority neighborhoods, bankruptcy and small firms' access to credit, and the effects of social capital, gender and race on the costs and availability of financial capital in mid-market banking.

A session on microenterprise lending investigated the role of lending in self-employment and how microcreditors use trust to reduce transaction costs and facilitate loan repayment. The final session at the conference examined the relationship between credit scoring and small business lending and the development and expansion of secondary markets for small business loans.

Proponents of credit scoring for small business loans offer several reasons why the technique may increase credit to LMI areas. First, credit scoring is less costly to both the borrower and the lender than traditional underwriting methods that require time-consuming review of credit reports and financial statements by loan officers. Second, because small business credit scorecards typically weight considerations about the character of a firm's principal, credit scoring could boost the amount of credit available to small businesses in LMI areas if business principals have good credit histories. Finally, credit scoring is a more objective method of underwriting and is thus less likely to produce illegal discrimination.

Those who are skeptical about credit scoring's effect on small business lending in LMI areas argue that scoring will limit the availability of credit. These skeptics assert that credit scores are likely to be unfair because LMI borrowers were underrepresented in the samples of past applicants used to construct the scoring models. If the data samples did not include significant numbers of LMI borrowers and LMI applicants have different characteristics than other borrowers, the scorecards may not accurately reflect the propensity of LMI applicants to repay their loans.

Critics of credit scoring are also concerned that, in assessing small business loans, banks will use scoring as a substitute for having relationships with the businesses. This practice could put LMI businesses at a disadvantage since these firms may rely on a relationship with a lender to overcome unfavorable perceptions of businesses in LMI areas. This relationship is especially important for firms that do not have formal business plans or audited financial statements typically used in traditional underwriting of commercial loans.

Atlanta Fed Study Looks at Scoring vs. Nonscoring

Recent research by Michael Padhi, Lynn Woosley and Aruna Srinivasan of the Federal Reserve Bank of Atlanta explores the impact of credit scoring on small business lending in metropolitan statistical areas (MSAs) in Alabama, Florida, Georgia, Louisiana, Mississippi and Tennessee. The researchers used a data set containing community, demographic, small business, small business loan, branch location and deposit information as well as the results of a survey on banks' use of credit scoring.

Their model controls for the influence of a number of factors, such as total businesses, housing units, lender branches, median income in a community and lender asset size. The researchers try to determine whether there is any significant difference in small business loan originations between low-income (less than 50 percent of MSA median household income), moderate-income (50 to 80 percent of MSA median household income), and high-income (greater than 120 percent of MSA median household income) census tracts for scoring and nonscoring institutions.

The Findings

The Atlanta Fed researchers' analysis finds that banks that use credit scoring do not lend significantly less to small businesses in low- or moderate-income areas than to those in high-income areas.

Controlling for community and bank characteristics, Padhi, Woosley and Srinivasan analyze data on the amount of small business loan originations of each survey respondent bank in communities with various income levels. Their analysis finds that banks that use credit scoring do not lend significantly less to small businesses in low- or moderate-income areas than to small businesses in high-income areas. Nonscoring banks, on the other hand, lend significantly less money to small businesses in low-income areas. The table shows the average difference between the small business loan dollars originated in low- and moderate-income areas vs. high-income areas. For example, if one controls for other relevant bank and community characteristics, a credit-scoring bank that lends $100,000 in a high-income community would lend $12,386 less in a low-income community, a difference that could occur by chance. A nonscoring bank would lend $51,164 less under the same circumstances. (It is important to note that the model shows an association only. The authors did not attempt to identify a causal relationship between credit scoring and lending patterns.)

Another important finding is that the presence of a bank branch in an area has no significant effect on the amount of small business loans made in that area by credit-scoring institutions. For institutions using traditional underwriting, however, branch presence has a strong positive impact on the quantity of lending in an area. This disparity may indicate that credit-scoring banks can make small business loans without a strong relationship with their borrowers and that nonscorers rely more on having a relationship with their small business customers.

Padhi, Woosley and Srinivasan conclude that credit-scoring banks tend to make small business loans equally across areas of all income levels, from low to high. In contrast, banks that do not credit score small business loans lend less to low- and middle-income areas relative to high-income areas. Furthermore, by reducing banks' reliance on branch presence for information gathering and business relationships, credit scoring may allow banks to lend more easily and with less costs to areas outside their branch coverage. Thus, credit scoring may increase the availability of small business credit or improve the terms on which that credit is offered by increasing the number of competitors lending in a given area.

The Effect of Tract Income on Small Business Lending
Average Loan Originations Scoring Lenders Nonscoring Lenders
Low-Income Minus High-Income -$12,386 -$51,164*
Moderate Income Minus High-Income -$9,923 -$33,595
Note: The table shows the average small business loan originations in tracts of a given income level minus the average small business loan originations in high-income tracts. Only the difference in originations between low-income and high-income tracts for nonscoring lenders is statistically significant. Census tract income data are from the U.S. Census Bureau.

*Significant at the 0.05 level