When consulting with our clients, one of the most frequent questions I get is how much impact does each variable have on the Intelliscore Plus Score?   Today, I will attempt to answer this question, without giving away our “secret sauce”. Variables affecting a commercial credit score, like Intelliscore Plus, are broken down into three key factors:

Recency – 50-60% of model weighting is focused on current payment status including the number of times delinquent, trade balance, and percent of accounts delinquent.   Another 5-10% of the weight is affected by the company age, industry risk associated with the SIC code, and the size of the business such as employee size.

Frequency – 10-15% of the model weight is focused in the number of derogatory items such as number of collections, liens, judgments and bankruptcies. In addition, another 5-10% is concerned with the historical payment patterns over time. Is the frequency of slow pay increasing or decreasing over time?

Monetary – Finally, 10-15% is weighted towards credit utilization. The ratio of account balance to recent high credit balance, the ratio of delinquent balances to credit limits, and balances carried in relation to the rest of the industry all effect the credit score.

By understanding what factors impact your commercial credit score, and the weights attached to them, you will be in a great position to manage your business profile. 

In a previous post I advanced the first of a two phased cross matrix approach to prioritizing commercial collections.  In this second post in a two part series, I’d like to expand on this idea.

Phase 2 – Collection Priority Intelliscore and Activity Age Indicator

In a previous post, I advanced the first of a two-phased cross matrix approach to prioritizing commercial collections.  In this second post in a two-part series, I’d like to expand on this idea.

Phase 2 – Collection Priority Intelliscore (CPI) and Activity Age Indicator

The activity age indicator cross matrix approach is an industry best practice introduced into the commercial collection industry by Experian.  In many cases the CPI, used in conjunction with bureau based attributes, will add further lift in collection prioritization.  Even though a business might have a high likelihood of collectability, it may, in fact, be out of business—or near out of business.

Experian uses a combination of BizAggs (business aggregates), called the Activity Age Indicator to evaluate if a business is active or inactive.  This activity-based approach is an industry best practice surrogate to determine if a company is in business or out of business.

The matrix then charts the distribution of activity, which creates a four-quadrant approach to collection priority:

• X-axis: CPI
• Y-axis:  Updates or credit inquires from 0-3 months, all the way to no updates and 11 years

The quadrants:

1. Accounts with a high probability of collection and high activity (meaning a business is still “alive” and not out of business) would be where you would want to focus collection efforts.

2. Low collectability-high activity accounts. Yes the business is still active, but it would have a lower likelihood of collection. Perhaps these businesses should be worked second.

3.  High collectability- low activity accounts. The business might score high in collectability but are really “dead” businesses.

4.  Low activity – low collectability accounts. This is where you would not want to focus. These businesses are most likely out of business and have a low predicted collectability.

In a previous post I advanced the idea of using a cross matrix approach to prioritizing commercial collections.  In this first post in a two part series, I’d like to expand on this idea.

Phase 1 – Portfolio Scoring with Commercial Collection Priority Intelliscore

The Collection Priority Intelliscore (“CPI”) was built to predict the likelihood of a successful recovery of dollars placed in collection on a business within a year of the collection.  It’s a score that we offer when scoring a portfolio. “Successful recovery” is defined as recovery of more than 50% of the amount placed for collection.  The score ranges from 1 to 100. High score indicates a high likelihood of a successful recovery.

By example, if you were to rank order the business going into collection according to the CPI score, and worked the top 10% of the accounts, the projected percentage of dollars successfully collected, out of the total dollars placed for collection, is 31%. Similarly, working the top 20% of the accounts results in 51% of total dollars recovered.

If you were to use a generic risk score for this purpose, the projected results are as follows: Working the top 10% of the accounts results in 12.6% of total dollars recovered, and working the top 20% of the accounts results in a 20.7% of total dollars recovered.

Therefore, it makes sense to use a score created specifically for collection purposes. The CPI provides a significant lift to you in predicting the likelihood of a successful recovery of dollars placed in collection on a business over and above a standard risk model.

Look for Phase 2 of the collection prioritization on my post next week…

One of the most frequently asked questions posed to me by clients is: Do I need a custom scorecard? To answer that question, let’s first start with the benefits of a custom scorecard (or model).  Here are a few of the best things about a custom scorecard:

1) Increased model performance. Typically a custom scorecard (or model) will generate a lift in KS (Kolmogorov-Smirnov statistic) from anywhere between 5% and 10%. And will vary between portfolios and industry.

2) Maximum flexibility in score range. A custom scorecard can be tailored to accommodate whatever score range you want. Do you want a score from 1 to 5, 1 to 10, or even 500 to 1000? A customized approach will accomplish this.

3) Manage and customize the minimum scoring criteria. What is the minimum amount of data necessary to generate a score? Does your credit policy call for strict criteria, or is a less conservative approach acceptable? These issues can be addressed specifically to your organization.

4) Custom scorecards can incorporate internal data specific to your organization. For example, application data can be used like financial statement data and loan specific data such as loan-to-value ratio’s and debt ratios. This type of internal data typically makes the model perform better.

5) Implementation can take place almost anywhere. Depending on what platform or system is being used, installation into your organization can be done with various degrees of ease.

So if these features of custom scorecards are interesting to you and your organization, then perhaps a custom scorecard is right for you!

In my very first post to this blog, I advanced the notion of “the credit risk trifecta” and how there are three general factors that contribute to a business credit score.  Well, I thought I’d follow that up with a sample of real data from Experian’s BizSource database.*

The following tables represent the mean values for three important (at a relatively high level) variables found to be more predictive of credit risk. For fun, I also thought it would be a good illustration to rank the top five highest scoring states (using the Intelliscore Plus model) and the lowest scoring states.

Keeping in mind that a higher Intelliscore Plus scores represents lower risk of delinquency, Table 1 are those states that rank in the Top 5 (lower risk).  Notice all five have relatively low Days Beyond Terms compared to the national average of 7 and almost 2X lower, on average, than the Bottom-5 in Table 2!

The Top-5 are also applying for credit less frequently (Alaska being the exception) compared the national average of 0.12, and their account balances are about $1,000 less, on average, than the national average of $5,500. All of which tends to raise the credit scores over the national average of 59.

On the other hand, looking at the Bottom-5, we see a horse-of-a-different color. Table 2 shows those states that rank in the Bottom 5 (higher risk).  Notice all five have relatively high days beyond terms compared to the national average of 7 and almost 2X higher, on average, than the Top-5 in Table 1!

The Bottom 5 are also applying for credit about as frequently as the national average of 0.12, and their account balances are over $1,000 higher, on average, than the national average of $5,500. All of which tend to lower the credit scores below the national average of 59.

*All data as of June 6^th, 2011

Given the current economic state of affairs, the need to develop an effective collection strategy has never been greater.  Unfortunately, the quantity of collection accounts has never been higher and the demand to prioritize which accounts get selected for additional recovery procedures becomes very difficult.  The good news is there are three industry best practices that can help.

First, a business that is no longer in business is probably not the best choice for recovery efforts.  No matter hard you try, you won’t get those businesses to pay. One way to determine if a business is still active is to evaluate the degree to which that business is receiving updates to their credit profile.  If a business is receiving frequent updates to their credit report (either by new trades being reported, additional inquires, or changing account balances) it is a very good indicator that the business is still active.  If active, then chances are good the business is open for recovery procedures.

Second, if a business is in bankruptcy, then collections efforts might be difficult. Given changes in the bankruptcy laws, certain lender liability issues are at risk with respect to various recovery strategies. Having access to a bankruptcy indicator from the credit report will give collection agencies and lenders a signal that these liability issues need to be addressed BEFORE attempting recovery efforts.

Third, collection recovery scores such as the Collection Priority Intelliscore are available to help predict the likelihood that amount placed for collection will be collected over a period of time.  This is extremely helpful in prioritizing recovery. For example, a business that scores ‘high” (a high probability of collection) should be placed first in the recovery queue because that business demonstrates the best odds of paying at least a portion of the amount placed for collection.

There are a number of factors that impact business credit risk scores.  Keep in mind that most risk models are built using multivariate statistical methods that not only look and each attribute but also look for the interaction between the attributes.  However, there are three general factors that will impact a business score.

Recency:  How recently has the business been delinquent?

Events that have happened recently tend to be most predictive of business behavior in the near future.  For example being days beyond credit terms (DBT) in the past 30, 60, and 90 days will tend to negatively impact, on average, a business’s credit score versus those that are current.

Frequency: How frequently is the business delinquent or applying for credit?

If a business has multiple beyond terms events then the algorithm will reflect this behavior and will tend to impact the score to the low side.  In addition, if a business is frequently applying for credit (called inquires) then this will also negatively impact the score.

Monetary/Usage: How large is the debt burden?

Business’s that carry large balances in relation to credit limits tend to be more risky than those who carry lower balances in relation to credit limits.  This is called the utilization ratio or balance-to-limit ratio.  As the debt burden increases interest payments also grow placing more stress on cash flows.  This tends to negatively impact a business’s risk score.