Rules Based Credit Scoring Methodology

CRF thanks Credit & Management Systems, Inc.  http://www.icmsglobal.com

 

Today’s credit professionals must make accurate, on-the-spot decisions. Of course, you have high value and/or high volume of credit decisions.  Departments are consolidating and slimming down as well as under more scrutiny from auditors.  Under these conditions, how do you perform the detailed and consistent analysis needed to avoid unneccessary credit risk?  Credit scoring can help. Credit scoring, by definition, is a method of evaluating the credit worthiness of your customers by using a formula or set of rules. Depending on the make up of your customer base, credit scoring can produce considerable benefits to some firms and somewhat lesser benefits to others.

 

The Benefits of Credit Scoring

 

 

 

 

 

 

 

The Different Types of Credit Scoring

Credit scoring is based on the assumption that past experience can be used as a guide in predicting credit worthiness. There are two types of credit scoring models.  Both can be statistically validated.

 

Judgmental Scoring Model

A judgmental scoring model is based on traditional standards of credit analysis. Factors such as payment history, bank and trade references, credit agency ratings and financial statement ratios are scored and weighted to produce an overall credit score.

 

The determination of which factors to use, and how each will be scored and weighted, is generally based on the credit executive's past experience with their company, the products or services they sell, and the industry they are in. Judgmental models are enhanced by comparing industry financial profiles using peer groups from (RMA) Risk Management Associates Statement Studies. Including scoring factors that reflect the individual characteristics and policies of their own firm further enhances the judgmental model.

 

Judgemental scoring is the most straightforward to implement because it uses your credit policies and decision process, the number of rules are easily set, and the grading scale can be simple or complex.  Therefore, it is easier to understand and augment.

 

Statistical Scoring Model

Statistical models function in much the same way as judgmental models. However, in choosing the factors to be scored and weighted they rely on statistical methods rather than the experience and judgment of a credit executive.

 

Statistical models consider many factors simultaneously, a process that calculates and analyzes multivariate correlation to identify the relevant tradeoffs among factors, and assigns statistically derived weights used in the model.  The key factors are generally captured from credit agency reports and the credit files of the client.

 

Statistical models are often described as a scorecard, a pooled scorecard, and a custom scorecard. A scorecard uses data from one firm. A pooled scorecard uses data from many firms. A custom scorecard blends a statistical model with some of the factors used in a judgmental model.

 

Who Uses Credit Scoring And What Do They Have In Common?

The first firms to use credit scoring were credit card companies and the consumer lending divisions of commercial banks. The huge number of transactions involved in consumer credit necessitated a computer generated score to approve and service their customers in a cost effective and timely fashion.

 

Fortunately, the credit information they needed for their statistical scorecards was readily available, much of it free. Their credit application provided data concerning employment, annual salary, home value, mortgage and other obligations. Additional data was available in consumer credit reports that were usually very comprehensive.

 

The information required to conduct a credit appraisal for a consumer is far less than that of a business. A salary of $50,000 can be measured easier than a bank reference or a financial statement.

 

Finally, the dollar amount of each consumer credit transaction is usually low, so that a single scorecard rejection of a sale will have little impact on overall sales revenue.

 

All of the above factors contributed to making credit scoring quite successful for those firms in the consumer credit business.

 

The second group to use credit scoring was banks, leasing companies and finance companies lending to small business firms. Similar to the credit card business, the numbers of transactions were high and the dollar amount on each relatively low. Since they could not afford to spend hours gathering and analyzing credit information on each transaction they turned to statistical scorecards.

 

Now, credit scoring is beginning to be used by business credit providers of all types.

 

As A Business Credit Provider Should You Use Credit Scoring?

If you have a very high number of customers, and the dollar amount involved with most of these customers is small, the decision is easy. In fact, you have little choice but to use credit scoring if you want to increase efficiency. Because these small business transactions are quite similar to consumer credit transactions, a combined business/owner credit-scoring model is very often used. By doing so, the credit executive can gain the same benefits that consumer credit providers enjoy; namely speed, accuracy, consistency and reduced personnel costs.

 

On the other hand, if you have fewer customers, and medium to large dollar amounts involved, you need to use credit scoring a little differently to gain similar benefits.

 

Consider first that credit scoring business customers is not as precise as credit scoring consumer customers because the amount of credit information varies considerably in business credit and the information is more difficult to validate and analyze. Secondly, the larger dollar amount of each transaction will have a far greater impact on overall sales revenue.

 

These differences can be compensated for by adjusting how you use the credit score. In consumer credit scoring, the ultimate goal is to say yes or no on 100% of the customers. In business credit scoring, you might want to set a goal of saying yes to the 80 or 90% of the customers whose credit score is good, and the dollar amount of the transaction is less than your credit limit model. In other words, say yes to the easy, safe transactions. Then, don't say "no" to the exceptions; instead refer them to a credit analyst to decide.

 

The credit information gathered for these exceptions will be considerably more extensive than for those customers that were automatically approved. Likewise, the credit analyst will typically have a sense of the relative importance of each piece of information gathered in the decision process and quickly reach a conclusion.

 

Perhaps the most important reason for business credit providers to use credit scoring is that scoring is the only effective tool to quantify the risk being taken. It is difficult for a credit executive to do a complete job without having scores to understand what risk is being taken at the portfolio level. It is even more difficult to affect changes in your firms credit risk policy unless you have something that allows you to quantify risk and communicate to senior management the fact that your accounts receivable portfolio value has changed. Credit scoring allows credit executives to fine tune credit risk guidelines over time.

 

An Example Judgmental Scoring Model

A judgmental scoring model is used in this example because it is an effective scoring model, and the simplest scoring model to implement. It uses your credit policies and decision processes, the number of rules are easily set, and the grading scale can be simple or complex.  And it does not require the services of a costly third party to create and maintain.

 

In this scoring model, the goal is to calculate an overall risk score based on the principle that the risk or credit worthiness of a customer can be evaluated on:

  1. Traditional credit information - This is credit information that is non-financial data that you would normally use when making a credit decision. Items such as control years, trade references and pay history fall into this category.
  2. Credit Agency information - The outside ratings that you normally consider in your credit decision process. The D&B rating, Paydex score and Experian Intelliscore fall into this category.
  3. Financial Statement scores – Ratios scored comparing peer companies within the same financial year by industry by size (sales or assets).

 

Each piece of credit information, from trade references to financial information, will be scored on a 1-6 scale in which 1 is the best score and 6 is the worst.

 

You establish the level of emphasis for each item by giving it a weighting. This is a subjective element that “customizes” the scoring model to your specifications.  Sample values for each category are shown below.

 

Traditional Credit Information Scoring:

 

Information

Weight

Score

Result

Pay history to others

0.15

3

0.45

Pay history to others

0.15

3

0.45

Bank rating

0.05

3

0.15

Trade reference #1

0.04

2

0.08

Trade reference #2

0.03

3

0.09

Trade reference #3

0.03

3

0.09

Industry credit group

0.05

3

0.15

Control years

0.10

5

0.50

NSF checks reported

0.05

6

0.30

Collection claims reported

0.05

1

0.05

Suits

0.10

1

0.10

Judgments/tax liens

0.20

1

0.20

Traditional Score

2.61

 

Sample Traditional Scoring Parameters:

Text Box: Control Years
1.	Over 15 years
2.	Over 10 years
3.	Over 5 years
4.	Less than 5 years
5.	Less than 3 years
6.	Less than 1 year
Text Box: Pay History to your Firm 
1.	Discounts
2.	Pays Promptly 
3.	Slow I to 15 Days 
4.	Slow 16 to 30 Days 
5.	Slow 31 to 60 Days 
6.	Slow Over 60 Days

 

 

 

 

 

 

 

As you can see in the sample provided, traditional items typically consist of non-financial information that you would normally use when making a credit decision. Feel free to add the items that you use in your particular industry.

 

These items might include some of the following:

 


Plant Capacity

Profit Margin

Country Risk

Annual Purchases

Number of Beds in a Hospital

Collateral

Occupancy Rate of Beds

Competitive Index



Credit Agency Information Scoring:

 

Information

Rating

Weight

Score

Result

D&B Rating

4A2

0.20

2

0.40

D&B Paydex

75

0.20

3

0.60

NACM Score

85

0.20

2

0.40

Experian days beyond terms

16

0.20

3

0.40

Experian Intelliscore

 

0.20

3

0.00

Credit Agency Score

2.61

Sample Scoring Parameters and their corresponding translation tables:

Text Box: Experian Days Beyond Terms
Zero to 10 Days Late = 1
Between 11--15 Days Late = 2
Between 16--20 Days Late = 3
Between 21--30 Days Late = 4
Between 31--40 Days Late = 5
Over 41 Days Late = 6

Text Box: Dun & Bradstreet Rating
4 A 1 = 1
4 A 2 = 2
4 A 3 = 4
4 A 4 = 6
3 A 1 = 1
3 A 2 = 2

 

 

 

 

 

 

 

The Credit Agency category will contain the third party ratings that you consider most important when making a credit decision. As in the traditional category, you will assign weights, in percentage form, to emphasize the importance of each rating. Also like the traditional category, items which are not scored are not used in the evaluation process--the items that are scored are simply re-weighted.

 

Also note since the scoring uses a common grade scale, understanding the multiple credit agency scores is easier to understand by eliminating the proprietary agencies alphabet soup of rating scores.

 

Financial Statement Scoring:

One ratio is not enough to analyze a financial statement. The best way to determine a firm's financial quality is to assess by comparing ratios of peers in three categories:

·         Liquidity

·         Profitability

·         Leverage

The 12 ratios listed below will provide a good measurement of the financial strength or weakness of a firm.

 

Liquidity Ratios

 

Profitability & Operating Ratios

 

Leverage & Coverage Ratios

 

In scoring a financial statement, the above 12 ratios will be computed and then scored through comparison to a published peer group such as the industry ratios published by Risk Management Associates in their Annual Statement Studies. Risk Management Associates (RMA), through their member banks, will receive financial statements throughout the year and then separate them into groups that are similar in the following ways:

 

RMA will then calculate and publish the upper quartile, the median and lower quartile ratios for each of 12 ratios. How they do this is reflected in the following hypothetical sample. The small sample size of 17 firms is used for illustration purposes only:

 

Peer group sample for Current Ratio


SIC -- 5083

YEAR -- 2003

SIZE – 3 (sales between $3-5 MM)

 

11.9

8.1

6.2

2.3

1.8 <-------- Upper Quartile

1.7

Right Arrow: Score1.6

1.5

1.4 <-------- Median

1.3

1.2

1.2

1.1 <-------- Lower Quartile

0.8

0.5

0.3

0.1

 

 

 

 

You will need to expand these three ratio levels in order to score on a scale of 1 to 6.

 

Current Ratio

Score

Above 2.2

1

Between 2.1 & 1.8

2

Between 1.8 & 1.4

3

Between 1.4 & 1.1

4

Between 1.1 & 0.8

5

Below 0.8

6

 


Each ratio is now scored on the 1 to 6 scale. The average score is determined for the liquidity, profitability, and leverage categories, and then multiplied by the weight assigned to each category, to arrive at the financial statement score.

 

 

Information

Weight

by category

Score

Result

Liquidity Ratios

 

 

 

Current Ratio

 

3

 

Quick Ratio

 

4

 

Sales / Receivables

 

2

 

Cost of Sales / Inventory

 

3

 

Cost of Sales / Payables

 

4

 

Sales/Working Capital

 

2

 

 

30%

18 / 6 = 3.00

0.90

Profitability Ratios

 

 

 

% Profit Before Taxes / Tangible Net Worth

 

3

 

% Profit Before Taxes / Total Assets

 

2

 

Sales / Total Assets

 

2

 

 

40%

7 / 3 = 2.33

0.93

Leverage Ratios

 

 

 

EBIT / Interest

 

1

 

Fixed Assets / Tangible Net Worth

 

3

 

Total Liabilities / Tangible Net Worth

 

2

 

 

30%

6 / 3 = 2.00

0.60

Financial Statement Score

2.43

 

Overall Risk Score

The overall risk score is developed by multiplying the traditional score, credit agency score, and financial statement score by the weight assigned. When a category score is not available the remaining scores are automatically re-weighted.  In the example below, if a financial statement had not been available, the traditional score would be weighted at 30/40's, or 75%, and the credit agency score would be weighted 10/40's, or 25%.

 


Category Weight Score Result

Information

Weight

Score

Result

Traditional score

0.30

2.61

0.78

Credit Agency score

0.10

2.49

0.25

Financial Statement score

0.60

2.43

1.46

Overall Risk Score

2.49

 


Overall Risk Score Scale

Highest Quality

1.00 to 1.83

Good Quality

1.84 to 2.66

Average

2.67 to 3.50

Below Average

3.51 to 4.34

Poor Risk

4.35 to 5.17

High Risk

5.18 to 6.00

 


As you can see in this example, the account is of “good quality” in terms of creditworthiness with a 2.49 Overall Risk Score.


 


Management Reports Made Possible By Credit Scoring

I. Portfolio Report

Historically, most credit executives used the Days Sales Outstanding (DSO) figure to measure the quality of their accounts receivable. This assumes, of course, that customer payment trends are related to risk. Actually, how a customer pays your firm is often a poor indicator of risk. Many high-risk customers pay promptly or within acceptable terms. Conversely, low risk customers often are given longer terms to accommodate special inventory programs.

 

A credit score that uses many types of credit information to evaluate a customer's risk is a better measurement than one single factor such as how that customer pays your firm. If this is true when evaluating a single customer, it is also true when evaluating all of your customers.

 

It follows then that a better measurement of the quality of your accounts receivable portfolio is to use the credit score for each customer along with the amount owed by that customer at the end of each month. By combining the total amount owed by customers in each risk category the credit executive will have a picture of how much of their portfolio is high quality, how much is high risk, and everything in between.  The following report is an example of portfolio analysis using credit scores:

 

Portfolio Analysis

Year

2000

2001

2002

A/R Balance

$20,087,185

$22,174,030

$25,108,033

High Quality

(1.00 - 1.83)

$2,812,001

14.00%

$1,980,004

8.93%

$1,081,003

4.31%

Good Quality

(1.84 - 2.67)

$5,429,002

27.03%

$5,588,005

25.02%

$4,750,004

18.92%

Average

(2.68 - 3.50)

$5,631,003

28.03%

$5,080,006

22.91%

$5,250,005

20.91%

Below Average

(3.51 - 4.33)

$6,215,179

30.94%

$7,326,007

22.04%

$9,010,006

35.88%

Poor Quality

(4.34 - 5.17)

 

$2,200,008

9.92%

$3,759,007

14.97%

High Risk

(5.18 - 6.00)

 

 

$1,258,008

5.01%

Percent

100.00%

100.00%

100.00%

 

This Portfolio Analysis illustrates a negative trend in the quality of the portfolio over a three-year period. In 2000 there were no Poor or High Risk customers, but in 2002 almost 20% of the customer base were considered Poor or High Risk. While the Portfolio Analysis uses an entire customer base, the credit executive may also want to segment their portfolio by product line, customer type or sales region to determine if certain groups of customers carry more risk than others.

 

 

II. Bad Debt Reserve Report

Using the portfolio analysis reflected above, and past bad debt experience for each risk category, the credit  executive can forecast an amount for the Bad Debt Reserve that is more accurate than basing it only on a percentage of projected sales.

 

Suppose that bad debts over the past several years had averaged 0.001% of annual sales. Using just that single variable, the Bad Debt Reserve for 2003 would be forecast at $270,000 or 0.001% of projected sales of $270 million.

 

However, if the credit executive forecasted the amount of bad debts by using projected accounts receivable and the past bad debt experience for each risk category, a much higher amount would be set aside.

 

The calculation for Bad Debt Reserve using this method would look as follows:

 

Bad Debt Reserve Forecast

Year

2002

2003

Bad Debt

% in the past

Bad Debt Reserve

A/R Balance

$25,108,033

$27,000,000

 

 

High Quality

(1.00 - 1.83)

$1,081,003

4.31%

$1,163,700

4.31%

0.0%

 

Good Quality

(1.84 - 2.67)

$4,750,004

18.92%

$5,108,400

18.92%

0.0%

 

Average

(2.68 - 3.50)

$5,250,005

20.91%

$5,645,700

20.91%

1.0%

$56,457

Below Average

(3.51 - 4.33)

$9,010,006

35.88%

$9,687,600

35.88%

2.0%

$193,752

Poor Quality

(4.34 - 5.17)

$3,759,007

14.97%

$4,041,900

14.97%

3.0%

$121,257

High Risk

(5.18 - 6.00)

$1,258,008

5.01%

$1,352,700

5.01%

4.0%

$54,108

Bad Debt Reserve Forecast

$425,574

 

The Bad Debt Reserve of $425,574 is probably more accurate than the $270,000 amount because it considers that the 2003 portfolio has a much higher percentage of Poor And High Risk customers than in 2000 and 2001.

Basing the bad debt reserve on past bad debt experience by category is somewhat similar to the computations by life insurance companies. While life insurance companies do not know which individuals in the 75 - 80 year old category will die next year, they can compute the percentage of the category very accurately.


 

 

Conclusion

Whether you decide to use credit scoring or not is an individual decision based on your customer composition and other circumstances of your own company. Which of the credit scoring methods you use, be it judgmental or statistical, is also an individual decision based on the perceived benefits to your company and the cost of implementation.  Rule-based scoring is straightforward to implement because it uses your credit policies and decision processes, the number of rules are easily set, and the grading scale can be simple or complex.

 

However, if you decide to implement credit scoring you will need computer processes that work with the scores. If credit scoring is being used manually, you will only have limited benefits. The advantages of speed, accuracy, audit tracking, and reduced personnel costs can only come if you add credit scoring logic and functionality into your existing computer systems or obtain them from a third party system.

 

 

 

About Credit & Management Systems, Inc.

Credit & Management Systems, Inc (CMS) headquartered in Lake Bluff, IL, USA, is a leading developer of comprehensive system solutions for corporate and commercial credit management. For more information on credit scoring and adaptability to your business, visit www.icmsglobal.com or contact CMS directly at either 847-735-9700 or sales@icmsglobal.com.

Copyright 1999 Credit Research Foundation
All material on this site, CRFONLINE.COM, is created or provided by CREDIT RESEARCH FOUNDATION, including text, graphics, logos, icons, and images, are the property of CREDIT RESEARCH FOUNDATION or its content providers, and are protected by United States and foreign intellectual property laws. The compilation of all the content on this Site is the exclusive property of CREDIT RESEARCH FOUNDATION and is also protected by United States and foreign intellectual property laws. You may download, view, copy, and print the materials on this Site for personal use only, provided that you do not remove or alter any trademark, service mark, or logo, or any copyright or other intellectual property notices. Except as provided above, you may not download, view, copy, print, reproduce, distribute, republish, display, post, transmit, or modify any material, or portion thereof, located on the Site in any form or by any means without the prior written consent of CREDIT RESEARCH FOUNDATION. CREDIT RESEARCH FOUNDATION reserves the right to revoke any of the rights granted in these Terms of Use at any time, and those rights automatically terminate if you violate any of these Terms of Use. Upon revocation or termination of such rights, you must destroy any digital or printed copies obtained from any portion of the Site. Unauthorized use of any material on the Site may violate copyright law, trademark law, and other laws of the United States and other jurisdictions. All rights not expressly granted in these Terms of Use are reserved.
 
CREDIT RESEARCH FOUNDATION and the CREDIT RESEARCH FOUNDATION logo are the trademarks of CREDIT RESEARCH FOUNDATION. The other trademarks, service marks, and logos used on the Site are trademarks of CREDIT RESEARCH FOUNDATION or others. Nothing on the Site shall be construed as granting, by implication, estoppel, or otherwise, any license or right to use any trademark without the prior written consent of CREDIT RESEARCH FOUNDATION. The designations ™, ®, ©, SM or any other intellectual property symbols reflect registration and/or use in the context of United States laws as they relate to such intellectual property symbols.