Whilst we transition into a new financial challenge for consumers, how can lenders learn from COVID-19 and amend scorecards to more accurately reflect consumer circumstances?

Portfolios will have been impacted to different degrees. What sample selection should you take? What needs to be considered from a performance outcome perspective? Have certain modelling variables changed?

This article by Experian’s Financial Services Consulting partner, Rebecca Galvin, highlights areas to consider when approaching scorecard and strategy updates and certain actions lenders could take.


Profile of customers attracted, taken on or impacted.

It doesn’t matter whether you’re building a scorecard for credit applications for a mobile phone, to manage existing customer lending in a bank or to assess whether to litigate for non-payment of an energy debt, you’re going to have to address a number of key questions.

Thinking about the individual profile of the applicants or customers affected will offer you a more sustainable approach. For example, if we consider originations decision making; how severe were the changes made over the pandemic and have the changes that have been made, here to stay? Lender approaches may have included:

  • Temporarily ceasing trading
  • Tighter policies e.g. increasing cut-offs / ceasing limit increases.

Whilst important for your scorecards to work well at the margins, if you don’t have sufficient data due to declining this population during Covid, you may need to avoid applications taken during this period.

The combination of changing policy and consumer behaviours resulted in a decrease in credit card balances by around a quarter during first 18 months of pandemic, with new lending down nearly a third in 2020 compared to the previous year and for personal loans, new lending was down 40%.

For revolving retail however, it was a positive story, driven by the availability of goods, with new lending up 11% in 2020 compared to 2019.

Did the profile of applications attracted during the Covid period change, for example, due to marketing, broker behaviour stemming from known policy changes or applicants attracted due to availability of goods to purchase? How different are they to the applications post covid?

Different solutions could be adopted dependent on the level of change seen:

If there is little change, you could sample as per the usual approach. However, if there is a shift in profile, then there’s a judgement on how significant the changes are and what you feel is indicative of the future. There are two main ways to do this:

  1. If the profile improved and contained extra perceived good payers, you could reweight down to look like the recent sample dependent on how different the population is.
  2. Alternatively, resample across a pre-covid period, assess and validate on the post-covid period, taking into consideration the change in profile.

Other typical sample considerations would also apply with sample selection, for example, seasonality, availability of historical data both from an internal and external perspective; bureau data covering variables trended over time.

You need to understand the impact on your portfolio, make allowances where you can, otherwise avoid certain time periods.

Portfolio Performance

Roll rates and vintage analysis become a must to understand what sample you need.

Analyse by those that took out an EPH and those that didn’t.

We are now getting to a point where there is sufficient post-Emergency Payment Holiday (EPH) performance to carry out analysis – Was the performance ‘x’ months after the EPH period vastly different? Some lenders consider using the EPH as a performance indicator and use within the good/bad definition.

When carrying out vintage analysis, splitting the population into EPH v non-EPH will help you understand how quickly the default performance is appearing

Consider altering good bad definitions to take into accountthe EPH data.

One approach could be to pause performance, for example, if the outcome period is 12 months and a customer was on an EPH for 6 months then you could look at the performance as at 18 months. Alternatively, you could keep the 12-month outcome but go for a softer good bad definition.

It is important to understand how your scorecards perform in pre, during and post-Covid periods, and consider extending the out of time analysis.

Changes impacting decisioning variables and future performance

The pandemic has affected people in different ways with some people suffering through loss of income and others benefitting through the ability to pay down balances as disposable income levels grew.

Furlough could have also had an impact as incomes were guaranteed at the same time as outgoings were reduced. Has this had an impact on making people look artificially good?

The tightening of policies and the reduced appetite for credit during the pandemic has impacted variables like the number of searches and balance to limits not working in the way that they would have historically. As people went out less and electronic payments were encouraged, the percentage of cash withdrawals reduced dramatically. Do balance and payment variables impacted during Covid need more scrutiny, would search trends also need this? For example, you may not wish to use cash withdrawals within your models. It’s not just a numerical exercise.

As we enter post pandemic life, changes in the credit world are still likely to cause problems. To date, the economy has remained resilient, but recent changes are going to impact people’s affordability considerably. The withdrawal of universal credit will impact some households, energy bills will double for many causing a major impact to the cost of living in addition to latest National Insurance increases. The current situation will also result in yet further impacts on the cost of living. This will almost certainly impact bad rates.

“I believe the biggest challenge facing lenders in the next 12 months will be how to adjust affordability assessments to keep pace with the change in cost of living”

This needs to be addressed to continue to meet regulation, for example, CONC 5.5A.18 (3) Where it is reasonably foreseeable that there is likely to be an increase in the borrower’s non-discretionary expenditure: during the term of the agreement; or in the case of an open-end agreement, during the likely duration of the credit (see CONC 5.5A.27R), which could have a material impact on affordability risk, the firm must take reasonable steps to estimate the amount of that increase.

Changes will be required throughout the customer lifecycle:

  • At point of origination; future proofing expenditure appropriately for each individual or household
  • In customer management to inform limit management or assess additional lending
  • At a portfolio level to understand the potential impact on arrears and to inform potential pre delinquency strategies
  • In collections when assessing income and expenditure

The importance of monitoring

It is strongly recommended that detailed scorecard monitoring is undertaken more frequently. Historically, it may have been sufficient to monitor the score, but characteristic monitoring is now important; and scenario tests could prove useful to keep on top of the changing environment we are living in.

Instead of a full redevelopment, an alternative approach could be to consider smaller refinements to current cards, more often.

Through systems such as Experian's Ascend Intelligence Services, we can help you monitor performance, recalibrate and test your models based on the latest insights.

Find out more

Summary of considerations

Considerations when approaching scorecard and strategy updates are as follows:

  • How old is the existing scorecard?
  • Would models and strategy benefit from new data?
  • Has the market changed?
  • Has the population changed?
  • Degree of impact to portfolio – adjusting time periods for scoring / review
  • How stable is the model?
  • What does our monitoring say?
  • Can I improve performance by using sophisticated modelling techniques?
  • How has behaviour changed and what impact has this had on data variables?
  • What strategy changes are needed? Additional policy rules, enhanced affordability assessment

Enhancing scoring and strategy involves careful consideration and business acumen in addition to good quality data and analytics. Experian Consultancy and Analytics can help you understand what the best option is for your portfolio.

To find out more about how consumers are reacting to the spiraling cost of living, and how lenders can work to support the changes in affordability – take a look at our Consumer ebook.