First, a bit of recent history: The Great Recession lasted 18 months until June 2009, the longest economic downturn since World War II.
The jobless rate doubled to 10.1%, and one-third of working adults faced unemployment. The mortgage default rate reached 10.1% in 2009, more than twice as high as the rate for any year since 1991. Since then, the economy has expanded somewhat; still, 7 million fewer people are working than before the recession.
So, the question is: Are you still using the same recovery models for collections that were built before the recession?
If so, or if you’ve only made minor changes, it’s time to look at some of the latest innovations in recovery modeling and analytics. In today's economy, it’s imperative to efficiently allocate resources to maximize dollars collected.
Whether your company is a very large financial institution or a small firm acquiring distressed debt portfolios, the interest in minimizing costs within collections operations has, undoubtedly, never been stronger.
There have been key innovations in recovery modeling that differ from traditional thinking. What we’ve determined is that blended collection-scoring models represent the most comprehensive way to segment generically and to set priorities for your collection portfolios. Blended scoring models are leading to:
• Improved scoring, segmentation and prioritization
• Reduced operational costs
• Maximized dollars collected rather than recovered accounts
As for traditional recovery methods, they now possess drawbacks and flaws. One major problem is that they treat a consumer account in, say, Danville, Ill., the same as one in Danville, Ky., or the Danvilles in California, Indiana, Pennsylvania and Virginia.
Collectors don’t take into account that the jobless rate in one Danville is twice as high as in another Danville. These accounts are called and treated the same but with different results.
This explains why leveraging geographic-based summarized data delivers insights into which accounts to place in your various priorities; contact methodologies; and settlement- and payment-arrangement terms, among others.
Yet, surprisingly, this data variable isn’t used en masse today. While you can’t use geographic or summarized data to grant credit or make credit decisions, it’s really useful for collections segmentation and strategy builds.
Here’s an example of the discrepancies you will find if you use such data: There are neighborhoods in Minneapolis, Minn., that have credit scores nearly 6% higher than the national average. In Indiana, there are ZIP + 4 ranges in some cities with twice the average number of accounts in collections than the national average.
Using geographic-based data should be just one of a variety of data sources you tap. By combining several different predictive data sources, from account-level performance data to credit data, you should gain significant lift over single-source data. You should find that dual-source data models perform better today and also adapt well to microtrends in the future.
Other key innovations have emerged in recovery modeling from the leading scoring providers as well.
One focuses on utilizing employment information that you gather from your customers. The more up-to-date information you can provide, the easier it is to build a capacity-to-pay scorecard on each customer. You can then take proactive steps even before delinquency to curb the impact of a consumer’s decreased capacity to pay.
Combine this data collection with other data you have already, and you’ll have a recovery model much more accurate than traditional models. Information such as payment dates, account activity dates and other available demographic information can be factored into the scorecard as well.
Also, customer or account segmentation is critical because the time-based “bucket” system of delinquency is not as effective as it once was. All accounts should be segmented by their capacity to pay and probability of collections or recovery.
In addition, collectors must align their treatment strategy with the current economic scenario. This will pay rich dividends if collectors employ aggressive settlement plans, nonstandard partial-payment arrangements and other flexible methods of term modification in continuing weak economic and high-unemployment periods instead of simply using business-as-usual practices.
Reflecting the difficulty today in collecting money because of the economy and high joblessness, there is a brand-new set of industry-specific scoring models — i.e., card, health care, utilities, telecommunications, etc. — segmented by age of debt and the level of credit balances.
This gives the user the ability to maximize the bottom-line return from their collection efforts, whether it is a company using the model for its own collection activity or for outsourcing to a collection agency. New developments in scoring models consist of two scores: the payer incidence score and the expected dollar score.
The payer incidence score determines the relative likelihood of receiving a payment during the six-month period after scoring. Thus, if account A has a higher payer incidence score than account B, then account A is more likely to make one or more payments during the period.
Alternatively, the expected dollar score provides the relative amount of expected dollars to be received during the six-month period following scoring. Therefore, if account A has a higher expected dollar score than account B, then account A is expected to pay more dollars during the period.
This collection-scoring model differs significantly from other available models, and the differences are compelling. The customer account-level data combined with credit data improves the likelihood that the right strategy will be applied to the right account, especially since the models are industry-specific.
In addition, hit rates are increased substantially since the account-level data and noncredit data spark a significant rise in the number of accounts that can be scored.
Because of its expected dollar score component, this recovery model provides a unique estimate of relative expected dollars to be collected, which previous generic scoring technology has not delivered.
Also, the expected dollar score can serve as the basis for portfolio account segmentation, simply by rank-ordering the accounts by expected collection dollars and segmenting the accounts at logical cutoff points.
New blended models offer the chance to allocate the bulk of your collection resources to the accounts with the highest potential for the most collection dollars.
You aren’t flying blind, and you will experience an increase in cash flow as a function of the effort applied. For a financial institution, it helps determine which accounts to keep for internal collection activity and which to turn over for agency activity or sell.
For an agency, it helps determine the accounts to work more intensely and which accounts to scale back the applied resources on. For both company and agency applications, it helps determine what level of effort should be applied and what costs should be incurred to maximize bottom-line results.
In determining whether this new unique set of industry-specific scoring models is appropriate for you, consider responses to these questions:
• Are you using an older-generation recovery score? • Are you happy with the performance of your current score? • Are you looking to improve your recovery dollars? • Are you attempting to reduce or control your collection costs?
• Are you analyzing treatment strategies for your collections accounts?
Your answers to this handful of questions will help you decide if you should pursue this new collections-recovery model.
Whatever you do on the recovery model front, remember that economic conditions and the individual situations of consumers have changed dramatically since the Great Recession began. So, if you haven’t changed your recovery models since then, now is undoubtedly the time to start.
David Ingram is the senior director of Collections Marketing at Experian.
