Strengthening Public and Private Credit Registry Ties

1
Strengthening Public and Private Credit Registry
Ties
World Bank 2004 In Beijing
By Larry Chiang
ã2004 UCMS, Inc. Confidential and Proprietary
Reproduction Prohibited
2
Public and Private SectorsWorking together in
harmony or discord?!

• Public credit registries (PCR) sharing of
  non-sufficient data amongst multiple agencies /
  branches of government
• Private sector sometimes fears losing competitive
  advantage if data is pooled with weaker
  companies
• No great models available to blueprint
  cooperation techniques
• PCRs dont share data to make data better

3

• Self-interest doesnt just govern, it RULES

4
Priming the Cooperation Pump

• Expecting parties to share goes against simple
  game theory
• What primes the pump? A government mandate, an
  olive branch, a philanthropic corporation, or
• A technical tool that works with EXISTING data

5
Jumpstarting A Countrys Credit System with Thin
File Analysis (TFA)
Incomplete Files (Private)
6
(No Transcript)
7
Thin File Analysis (TFA) Leverages

• TFA gives private credit registries a technique
  without full cooperation of PCRs (Public Credit
  Registries)
• TFA allows for a first stab at credit
  underwriting
• TFA is dynamic and adjusts to the level of
  cooperation (initially low interdependence, later
  more cooperation)
• TFA helps a developing nation import a mature
  nations credit scoring system without symptoms
  and side effects

8
A Quantum Leap Forward With Thin File Analysis
Old Way (Bad Chicken, Bad Egg) Thin File Analysis
Who Private and public sectors are at odds Together
Where Your server Your server
What Scorecard unknown Formula FICO Development
When Not really even once Underwritten 3x
Why Stab in the Dark Scientific!
How Copy from other country and hope for non-collapse Custom Country Optimization
9
Origins of TFA

• Thin file analysis was used where public
  information was there but not organized
• Private registries (Experian Equifax Trans Union)
  had thin credit files (i.e. very little credit
  data)
• TFA bridged the gap
• TFA catalyzed underwriting and credit line scoring

10
The Old Way

f (a, b, g)
The old technique of measuring credit worthiness
with a score from 350-800. This (with all its
parameters) is a one dimensional match.
11
When the Old Way Works Great

• In countries with a mature credit industry
• Great with fully stocked public data compilation
  that is organized and accessible. Public
  data by its nature is difficult to access,
  process and underwrite
• Its great when the credit industry is ONE BIG
  COOKIE not when its crumbs youre piecing
  together.

12
Fighting Over Crumbs

• Mature credit industry countries dont bicker
  over who controls credit data
• Burgeoning credit countries wrangle over control
  of data
• Thin file analysis primes the pump when credit
  files are thin and morsels of public record data
  can scientifically build up a credit system

13
Adding a New Axis- Public Registries
f is the function with traditional credit data
with CRAs e function factors in applicant data,
govt agency data more
14
Thin File Analysis Marries Both
f is the function with traditional credit
data e function factors in applicant data,
govt agencies more
15
The New Way

You set how much influence the epsilon function
gets
16
Recipe for a Rough Transition

fusa (a, b, g)
fkorea(a, b, g)
The credit scoring system is directly imported
from one country to Another. Charge-offs shoot
over 17
17
Why direct transferal doesnt work.

• Cultural differences in perception of credit
  default
• Different realities of credit default
• Behavior predictions based on income to asset
  ratio dont transfer country to country
• Private credit registries wield different
  influences

18
Customize an Imported Credit System
f is the function with traditional credit
data e function factors in applicant data,
govt agencies more
19
Sequence of Steps to aCountry Specific System

• (2) is goal for China in the goal for 2007
• Is the goal for 2011
• (4) Is the goal for 2015

20
Private Registries And- Public Registries
f is the PRIVATE function that slightly depends
on e function e is the PUBLIC function
supports a temporarily weak f function
21
Perhaps Yet Another Axis is Needed