How to use Tilt-by-Risk Principles to Design Effective Collections Strategies
Scorecards are the gold standard of retail credit risk assessment.
So, you’ve got a scorecard! 🎉🤩🥳
The really great news is that getting and implementing the scorecard is the difficult part. The less fantastic news is that a scorecard is not a strategy.
I thought I would outline what a Collections Strategy might look like, once you have a scorecard in place.
But before I dig into designing collections strategies, I thought I would chat a little bit about Behaviour Scores versus Collections Scores.
On one level, any scorecard that makes use of behavioural information, as opposed to demographics, social or other data, is a behavioural scorecard. In the credit risk world, though, we refer to scorecards that have an outcome window of 6 months as behaviour scorecards. Collections scorecards use similar variables to behaviour scorecards; however, they’re predicting the likelihood that a qualifying payment will be made on the account this month and that it will not become increasingly delinquent.
And what do I mean by a qualifying payment? This is a payment that is sufficient to prevent the account from rolling into a higher level of arrears. It does not mean that they pay the full arrears, although, if that is what you wanted to predict, then, of course, you could build a scorecard designed to do that.
Collections strategy design
Not all accounts that are in arrears are the same. Some accounts have missed more payments than others; some have higher balances; some have a promise to pay in effect; others may have recently broken their promise to pay. Some accounts may be experiencing hardship. Some accounts may have been outsourced to external collections agencies.
A collections strategy needs to cater for all these eventualities.
Good collections strategies identify accounts that are similar from a policy point of view. For example, Hardship accounts. These accounts are usually carved out from the population for special treatment. Similarly, accounts that are identified as Skip/Trace accounts – accounts where we have not been able to make contact with the customer and need to do something differently are typically queued separately for specialised treatment.
We also know that someone who is one cycle in arrears is generally easier to action than someone who is three cycles in arrears. And we don’t want to action accounts that have been passed to an external collections agency.
So, taking all these, and other factors into consideration – how do we go about building an effective collections strategy?
Identify accounts that need specific treatment
The very first thing I like to do is identify all those accounts that ought to be treated in a specific way because of their specific circumstances.
We typically identify these accounts through Block Codes or Status Codes. Typical examples include:
Hardship – the account has met the Hardship threshold
Administration or Garnishee Order – the account is under administration and very specific agreements have been reached (typically through the courts)
Fraud (under investigation for fraudulent activity)
Deceased – the account holder is deceased
Insolvency or Bankruptcy
Jailed – the account holder is in prison
Legal – there are legal actions being taken on the account
External Debt Collections Agency – the account has been handed over to an external agency. In some organisations, this activity is performed in-house by a speciality team.
Skip/Trace – these are accounts that we have not been able to successfully make contact with. Each organisation will have their own threshold, but at some point, it may be necessary to identify accounts that we cannot reach and see whether we can find updated contact details.
Once all these accounts have been identified and routed to their respective queues for the appropriate actions, we still have a pile of accounts to action.
At this point, I like to segment accounts based on Delinquency.
A question worth asking is whether you want to action pre-delinquent accounts. Historically, accounts that were not in arrears were not actioned. This was simply a financial decision – when the only action options were to phone the customer or send them a letter, and both those activities were fairly expensive, financially, it did not stack up. With the widespread adoption of text and email, customers who are not in arrears can be cost-effectively actioned.
Let’s assume we do want to action pre-delinquent customers. This means our next decision key is delinquency. And in order to ensure that we action pre-delinquency accounts, it means we include all accounts that are up to date with a payment due in our 'pile' of accounts.
What about other factors?
In addition to delinquency, we might be especially concerned about accounts with high balances or high utilisation. And if we have a collections score, a behaviour score or a bureau score, we should leverage all that information too.
Based on the above, we now have the decision keys that we will use in our Collections Strategy:
· Block or Status Code
· Balance Outstanding or utilisation
· Risk (as indicated by score)
Tilt by Risk
Tilt by risk is a concept used in credit risk; think of it a bit like a see-saw, as credit risk scores increase, indicating lower credit risk, we decelerate actions; conversely, as credit risk scores decrease, indicating higher credit risk, we accelerate actions.
The diagram above conceptually how this works. And in the diagrams below detailing a simple collections strategy, you can see how this is used in practice.
An Early Arrears Collections Strategy
In this early arrears collections strategy, we have first identified all accounts that qualify for specific treatment based on their status code or block code. All those accounts are routed to the appropriate queues and teams.
Before we start building out a strategy, it is helpful to think about the kinds of groups of customers we may have, and how we want to treat them. There is little point in creating a segmentation tree and then treating everyone the same. Similarly, if your operations team only has enough resources to call 60% of your delinquent accounts, you want to ensure that they are being as effective as they possibly can be by ensuring that higher-risk accounts are called, and lower-risk accounts are either sent an appropriate text message or left to self cure. (In the event that these accounts do not self-correct, they will be identified as high enough risk to qualify for appropriate action.)
In our example strategy shown below, our second decision key is Delinquency, and our third decision key is Balance Outstanding. For both up-to-date accounts and accounts that are one cycle in arrears, we have identified accounts that have high, medium or low balances. Accounts with higher balances have higher repayment amounts, and therefore, as arrears on these accounts increase, it becomes increasingly difficult for customers to catch up on missed payments.
Our final decision key in this strategy is Credit Risk. This is typically identified through a behaviour, collections or bureau score, although a Risk Profile could also be used. It’s worth noting that I have seen internal (behaviour or collections) scores used together with external (bureau) scores to great effect.
Now, let’s work through sections of the Collections Strategy to understand the relative risk posed by various groups of accounts
In the image below, consider the accounts that can be identified with the coloured cells. We’ve identified accounts that are up to date with high balances, and we can see that they are medium to high credit risk. These accounts are still in order, but, due to underlying behavioural patterns, may fail to make a qualifying payment this month.
So, we have decided that we want to take action to ensure that they make their payment. We also need to be mindful that these customers are still in good order, and so we ought to be mindful of what type of action we take. We can see that these customers have a low risk relative risk profile, when compared to the other accounts. Therefore, we’ve decided that a very gentle customer service type text message is appropriate. This text message is identified as Text1.
Contrast the above with the image below, where we have identified our highest relative risk customers. They’re in arrears, with high balances and have high credit risk, typically indicated by low scores. Comparatively, they are very high risk. Therefore we’re going to put them in a queue for a call. The tone of this call is aligned with the risk that these customers may fail to make their payment obligations and may highlight consequences of continued missed payments.
As you can see, even with this very simple and illustrative collections strategy, we use a combination of decision keys, including the account score, to estimate the relative risk of accounts. And once we have done this, we treat accounts based on their relative risk profile. This ensures that we treat customers appropriately and optimise the use of our collections agents.