The Data Analytics Blog

Our news and views relating to Data Analytics, Big Data, Machine Learning, and the world of Credit.

All Posts

10 ways the COVID-19 crisis will affect your credit models (PART 1)

April 5, 2020 at 12:25 PM

Black Swan

One of the basic principles of credit scoring and modelling is that the “future is like the past”.  Whilst robust credit models may be calibrated on multiple time periods, this assumes that trends in the past represent what is going on today.  COVID-19 is a black swan event – meaning in the modern day it really is unprecedented.  If you have never come across the term black swan, or if you have but no idea the origin, I recommend taking two minutes to read its really interesting etymology.

Over the last 20 years we have witnessed hundreds of our clients move from manual and rudimentary strategies to sophisticated, model-driven and automated ones. This had occurred across the customer lifecycle from acquisitions (hunting for new customers), originations (on-boarding), to account managing, financial models including IFRS9 provisions, and finally collections and recoveries. 

At Principa we live and breathe models. In this blog we list 10 key ways how we believe COVID-19 will affect your credit models. It’s in two parts, the first part deals with scoring across the lifecycle. The 10 items are as follows:


  1. Bureau scores
  2. Acquisitions scores
  3. Application scores
  4. Behavioural scores
  5. Collection scores


  1. Payment holidays
  2. IFRS9
  3. Sales and budgeting models
  4. SMEs
  5. Vintages

In a future blog we will cover some remedial actions and dive deeper into some of these topics. 

We’ll be discussing the contents of the blog in a series of videos.  To receive them you can subscribe to our YouTube channel or you can sign up here. 

If you would like to chat to us to understand how we might be able to help you with your pressing concerns, please contact us here.

  • Credit Bureau scores drop

Credit scores are used across the credit lifecycle, but most notably in originations. As more individuals miss payments, so their credit score will drop.  Does this represent a heightened risk or is this temporary effect due to the COVID-19 crisis? Does the score calibration represent the actual risk I should expect 12 months from now?  It should be noted that the utilisation of payment holidays will not affect variables utilising amount paid or percentage paid but will affect variables that utilise a worsening of cycle delinquent status (the latter variables are frequently used in bureau scores).  

In summary we expect credit bureau scores to drop as the overall risk of the population does drop, but not necessarily in line with the expected score-to-odds relationship as per the pre-COVID19 calibration.  We also expect some higher risk customers to benefit from the payment holidays which may flatten out the score-to-odds relationship.

We discuss this further in a VLOG that you can view here.
  • Acquisition: Risk and Response models

Along with the COVID-19 crisis, South Africa has also experienced two credit agency downgrades. These have all bought much stress to the South African consumer and will affect the taking up of credit offers.

Acquisition models typically comprise a response model (to a campaign) along with a risk model.  For insurance early attrition models may also be used. Just as the bureau scores and application scores may underestimate risk, so the risk acquisition models may do the same initially.  Over time we’d expect the score distributions to deteriorate too.  For response models it very much depends on the product. There may be an increased interest in unsecured loans as consumers experience more economic stress: so, response models may under estimate response to unsecured credit campaigns. Insurance, apart from health insurance or income protection may drop in response. Ultimately these models will need to be recalibrate (or preferably rebuilt) to deal with the new-normal post COVID-19.

  • Application scorecards

Application scorecards will also be affected by the COVID-19 crisis.  Whilst demographic data may not change, the score-to-odds ratios may indeed change, and the scorecards will ultimately need a “down-turn” calibration.  Self-employed individuals may be at higher risk than normal. Scorecards will likely “flatten out” – i.e. may not rank-order risk in the same way. Therefore, aggressive risk-based-pricing strategies may not be as prudent as the underlying risk at each scoreband/risk-grade may not be the same as previously expected.  If you are not monitoring or only monitoring your scorecards annually/biannually then we’d recommend you make this a monthly activity during this crisis and address any aggressive risk-based-pricing strategy.

  • Behaviour scores dropping for the highly stressed population

Well-made behaviour scores should have been built utilising data through the credit cycle.  However, Corona-type shocks do not exist in recent historic data. It should therefore be expected that behavioural scores are likely to drop for a considerable number of customers over the next twelve months at least. Payment behavioural patterns will also be impacted with payment holidays and the scores might flatten out (with otherwise low scoring customers scoring a little higher).  Regular (monthly or even weekly monitoring) and more conservative strategies are recommended.

  • Collection models

Collections models are likely to be affected on a variety of fronts. Some of the key changes will be the following:

  1. Severe operational changes during the COVID-19 lockdown will likely see higher roll-rates than expected as fewer delinquent customers are contacted. The models will therefore over-estimate probability of payment or under predict probability of rolling.
  2. Stressed population will likely see a lower propensity-to-pay or higher probability of default than what the models originally predict.
  • Payment projection scores for customers in late stage may underestimate total recovered amount for accounts particularly with a sustained period of hardship and lack of contact during the initial phases
  1. Right-person-connect and right-time-to-call models will predict differently during the lock-down phase as customers are staying at home.
  2. Preferred instalment models may over-estimate affordable instalment as customers’ cashflows are severely affected during this period.
  3. Optimal settlement models may over-estimate affordable settlement amounts and this is likely to continue during the sustained period of hardship expected.

Some models may require redevelopment, some a recalibration and some (like connect models) may experience a short period of inaccuracy but will then rebound to normal. ALL MODELS SHOULD BE MONITORED MONTHLY.


Whilst black swan implies an event, we believe that the COVID-19 crisis will be long-lasting, and the effects felt long after the crisis has abated. We would strongly recommend as a lender you consider addressing these effects by assessing them and taking remedial action.  If you would like to chat to us to understand how we might be able to help you with your pressing concerns, please contact us here.


Contact Us to Discuss Your data analytics Business Requirements

Thomas Maydon
Thomas Maydon
Thomas Maydon is the Head of Credit Solutions at Principa. With over 17 years of experience in the Southern African, West African and Middle Eastern retail credit markets, Tom has primarily been involved in consulting, analytics, credit bureau and predictive modelling services. He has experience in all aspects of the credit life cycle (in multiple industries) including intelligent prospecting, originations, strategy simulation, affordability analysis, behavioural modelling, pricing analysis, collections processes, and provisions (including Basel II) and profitability calculations.

Latest Posts

Solving the Credit Unaware Challenge with Psychometrics

At Principa, we engage with clients and organisations across the entire credit lifecycle and track the focus of the South African credit industry. For nearly ten years the focus has consistently been in the collection space, but recently (since early 2021) this has changed and a large number of our clients are focused on acquisitions and originations.

Predicting Customer Behaviour (PART 2)

In Part One of this two-part blog, we started providing a short overview of just some of the propensity models that Principa has developed. In this Part Two, we continue to look at different types of propensity models available across the customer engagement lifecycle that are used to predict behaviour and solve business problems. 

PART 2: How to Cure the Post Pandemic “Collections” Symptoms

In PART 1 of this two-part series, we explored how the current socio-economic climate resulting from the lingering financial hangover caused by the pandemic is negatively impacting the consumer's ability to settle a debt.