Account Management Score vs Collections Score by vivi07

VIEWS: 7 PAGES: 32

									Scoring for Distressed Debt Buyers and Sellers
Richard Yap VP, Business & Product Development SCORE STATISTICAL CONSULTING

Overview of Scoring

Account Management Score vs. Collections Score vs. Debt Score

Front End
Application scores try to identify a small number of bad accounts among the many good

Back End
Collection scores try to identify a small number of good accounts among the many bad

Sale
Dormant Debt scores are used as a tool to assist in the evaluation of portfolios for sale

Account Management Score vs. Collections Score vs. Debt Score

Understanding Scoring
• There are three instances whereby a score cannot be generated: • An “edit/reject” results when a file was incorrectly formatted or incomplete (e.g. business name) • A “no hit” results when an accountholder could not be found at the credit bureau • A “no score” results when an accountholder could be found at the bureau, but their file was too thin to generate a score (e.g. new/first credit card for an accountholder) • The reason behind the absence of a score can often provide insight regarding the quality of these accounts

What Do Scores Really Reflect?

Challenges with Credit Bureau Based Debt Sale Models
• Availability of reliable data – is what happened in the past indicative of the present or the future? • Credit bureau reporting R9/I9 tradelines • Many types of portfolios for sale, all of which have different compositions, ages, industries, histories etc. • There are a host of assumptions built into the sample design – i.e. how does one compose a “representative” sample?

Potential Challenges with Using Historical Collection Curves
• Use of historical collection curves for “like” portfolio comparisons e.g. card “A” 12% liquidation over 36 months; card “B” is a “similar” portfolio and should return close to the same

Potential Challenges with Using Historical Collection Curves
• Will what happened in the past be indicative of the present or the future?

• • • • •

Variables could include: Credit grantor adjudication/acquisition strategies Credit grantor product mix Account management strategies Collections processes (internal and/or external) Average sized balances etc.

•Dollars Collected For Segment X:Segment Y

•1.00:2.00 (original)
•1.00:0.66 (sale) •Material change in the relative collectability of Segment X vs. Segment Y

Benchmarking

What is Benchmarking?
• Benchmarking scores is used to compare shifts or movement between score distributions • Using scoring consistently, qualitative and quantitative observations can be drawn that provide insight into the direction and velocity of trends For example, are a bank’s credit cards improving, steady, or degrading, and if there is a trend, what are its characteristics? (e.g. is it rapid, slow, cyclical etc.) • An important component to benchmarking is capturing the data required to perform the analysis in a consistent and complete manner

Benchmarking – Some Examples
Benchmarking Yourself - observing shifts in score distributions for a particular portfolio of accounts over time
–e.g. 1 Is the quality of Card A applicants today better or worse than 2 years ago? –e.g. 2 Purchasing debt on a forward flow contract, is the quality of paper being purchased improving or degrading over time?

Benchmarking Against Others - comparing the score distributions of an organization versus that of its industry
–e.g. 1 How does the quality of Bank A’s unsecured lines of credit compare to those of the rest of the industry? –e.g. 2 What is the average score for our retail financing applicants versus the average score for the industry?

Benchmarking Others - assigning like business to evaluate the performance of other organizations
-e.g. 1 Retailer A early outsource program for low scoring accounts –which of 3 agencies performs best on this business? -e.g. 2 Using scores as a tool for segmenting business for champion/ challenger comparisons in risk-based pricing

Pre-Scoring Portfolios for Sale

Pre-Scoring for the Seller
• From a seller’s perspective, pre-scoring accounts prior to sale can be advantageous to:
– Have an estimate of portfolio value prior to sale (e.g. establish reserves, estimate of “fair” price etc.) – Scoring a portfolio and retaining highest quality accounts will obviously impair the value of the portfolio

Collection Curve – Card A

1.32% (months 19-38)

Portfolio Comparison Card A vs. Card B
Cumulative Score Distributions at 18 Months Delinquent

Pre-Scoring Accounts
• As part of a portfolio sale package, potential purchasers typically receive a file from the seller that includes accountholder information • This information could be used by the purchaser to pre-score accounts to assist in analyzing the potential returns for a portfolio • This portfolio review involves score distributions, balances sizes, and a multitude of other empirical (or intuitive) factors

Pre-Scoring Accounts con’d
• Pre-scoring could involve sending the entire population of accounts available for sale or just a “representative” sample of accounts • Using scoring with other evaluation criteria (e.g. balances, date of last payment, geographic segmentations etc.) can provide a means to examine:
– The portfolio in question as an independent exercise – Compare against similar paper purchased and scored (especially useful for debt purchased on a forward-flow contract)

Pre-Scoring Example

•All things being equal, it would appear that Portfolio A (pink) is the more attractive because the score distribution is higher overall relative to Portfolio B (blue)
•Average scores can also be examined

Expected Return Calculation

Pre-Scoring - Balance Size Consideration -

+ 498,610

+ 440,080

Segmentation Using Scoring
• Debt sale scores are difficult to build • Employ scores with additional information from the credit grantor and/or agencies to maximize the overall return • Information could be geographic, demographic, product-based, application-based, behavioral etc.

Segmentation Example
• • • • • Averaged-sized balances Geographic segmentation Date of birth Charge-off date Time on books

Forward-Flow Contracts

Forward-Flow Contracts – Purchasing Debt
• Forward-flow contract: an agreement for a specified term by which a seller agrees to send written off debt to a purchaser on a routinized process as soon as it becomes available • For paper purchased on a forward-flow contract, scoring is advantageous to: – Immediately determine whether there exists material volatility in the quality of accounts being forwarded monthly (i.e. don’t have to wait for collections outcomes) – Use scores to assist in prioritizing accounts and drive collections workplans/strategies

Volatility Analysis - Historical Score Distributions -

Volatility Analysis - Average Scores • • • • • Average Score Month 1: 679 Average Score Month 2: 645 Average Score Month 3: 574 Average Score Month 4 :520 Average Score Month 5: 415

Volatility Analysis - Population Stability Index • This index is used to determine whether there is a statistically significant shift in the score distributions of any two populations • e.g. pay vs. no pay, cure vs. no cure, score distribution in Month 1 vs. score distribution in Month 5 etc. • an index of 0.10 or less is indicative of no real change between the populations • an index between 0.10 and 0.25 indicates some shift • an index greater than 0.25 signifies a definite population change • This method goes beyond examining only average scores since it assists in identifying anomalies within or between the score distribution

Calculating Population Stability Index

Forward-Flow Strategies
• Keep the portfolio
– As per collections scoring, prioritize accounts and use scoring as a major tool in developing work plans

• Use cutoffs and “cream” the accounts
– Work the most lucrative accounts in-house for collections, and sell lower scoring accounts

• Much depends on whether there exists a sufficient volume of accounts to segment • This strategy applies to both buyers and sellers

Questions?
Richard Yap richard@scorestat.com 416.861.1417


								
To top