What the Papers Proposes to Do

1
(No Transcript)
2
What the Papers Proposes to Do

• Objective determine the feasibility of a
  pooled data SME credit scoring tool in Latin
  American countries
• Not intended to be a business proposition (not a
  functioning product)
• But Fair Isaac stands to gain an advantage. How
  would that be handled?
• Hoped-for implication broadening of access to
  credit by lowering the costs of screening and
  gauging default probability
• Underpinnings
• Scoring models (pooled data) are working well in
  the U.S. since the mid-1990s, reducing importance
  of relationship lending in case of small firms
• Information on the business owner is a strong
  predictor of the default risk of a small
  enterprise (firm opacity can be overcome via data
  on owner)
• Method
• Pool data from multiple banks (and credit
  bureaus?) in 3 Latin countries
• Develop a generic score card mapping
  characteristics to probability of default
• Test validity against each lenders portfolio
• Results pending

3
Limits to Commoditization of SME Lending

• Heterogeneity of SMEs
• Heterogeneity limits standardization and
  applicability of law of large numbers
• Difference with consumer, mortgage, and micro
  loans may be of nature, not degree
• Is there a size threshold?
• Key dependency predictive power of information
  on business owner suggests so
• Implication loss of predictive power as SME
  grows (but remains informationally opaque) and
  linkage between ownership and management weakens
• Are we then not just talking about the larger
  firms among micro enterprises?
• Relationship lending goes beyond screening much
  is about monitoring
• Scoring methods can facilitate screening but not
  necessarily monitoring
• Mid-size enterprises are still opaque and require
  intense monitoring
• E.g. Chilean banks mix customized scoring with
  strong monitoring policies
• Scoring may not reduce costs in SMEs lending as
  much it does in high-volume consumer and micro
  loans, or in well collateralized and standardized
  mortgages
• Limits to transferability from USA to the
  tropics, and within the tropics
• Major differences in systemic risks macro
  volatility, weak contract enforcement
• Issues for pooling data across Latin countries
• Contract enforcement matters more as loan size
  increases ? scoring, even if it worked, seems far
  from being a silver bullet