These are challenging times, much worse than the 2008 crisis. The impact of COVID-19 has brought the world to a standstill. While the aftermath appears similar in both cases, the reality is different. The 2008 economic crisis was a slow event which built up over a period of time. The current scenario is an unforeseen event where immediate global lockdowns have created a dip in the economy

Unlike the banking collapse of 2008, the 2020 pandemic has muted demand impacting consumption of goods and services across industries. We’ve seen small businesses shutting down and an increase in layoffs. This can have a potential bearing on an individual’s cash flow and the ability to pay off their dues.

While the primary focus of governments across the globe is to protect the health and safety of citizens, they are also taking steps to safeguard the interest of the consumer. CARES Act in the US covers direct payments to families impacted by the pandemic, benefits to unemployed individuals, and grants for businesses that have suffered.

Governments across the globe have also stepped by providing short term moratoriums and payment breaks to their citizens.

Impact of COVID on the banking system
With growing job losses, slower demand and sales, and declining revenues, borrowers and banking customers will soon start seeking extended financial assistance. This will further stress bank assets and might require extreme measures for recovery. As a result, banks and financial institutions are going to face multifold challenges in short to mid-term basis as below:

  • Increase in losses due to sudden spike in delinquencies and loan defaults
  • Revenue compression impacted by an individual’s ability to borrow further in these challenging times
  • Cost optimization in every department without impacting the output

Challenging times ahead for debt collections
The collections teams within banks are facing a different challenge altogether. Federal and state agencies in different countries have implemented restrictions on debt collection in the short term. Once COVID-19 deferral ends, banks are expecting delinquencies to rise and will need to handle the situation with the available workforce.

Borrowers with no history in delinquencies will start entering collections bucket due to COVID-19. It will be essential to identify these customers from perennial defaulters and treat them separately.

Post moratorium, contact centers will have their hands full with an increase in outbound call volumes. Collections manager will have the task of managing the workloads for their teams.

Collections team have been leveraging risk models in the past to identify the risk of their delinquent borrowers and suitably define contact strategies. There is a high chance that these models will under- estimate the actual defaults and banks relying on such models could face higher losses.

RELATED ARTICLES

Restructuring and managing lending and debt collection strategies in the time of COVID-19

How a leading US based bank used Predictive Insights to lower delinquencies and net loss

Why current banking risk models are not enough
Banks use risk models to predict and identify the risk of a customer.

One of the major regulatory requirements for banks is to provision for their credit losses as per IFRS 9 & CECL standards through a forward-looking Expected Credit Loss (ECL) model, which primarily looks at the borrower’s probability to default (PD). With the current pandemic, risk managers within the banks have their task cut out and will need to re- think their approach.

Traditional risk models have been built on historical data to predict a borrower’s propensity to pay. Risk modelers typically use historical data to build these models. While these models have been performing strongly in a steady economy, they could perform poorly in the current crisis due to the following reasons:

Unprecedented event:

Given that COVID-19 is an unprecedented event, historical data in banks will not reflect the current crisis or anything similar to what the model can learn from.

Traditional models within banks look at limited financial attributes of a borrower like historical payment patterns and credit scores to risk segment borrowers. In the current crisis, there may be a scenario where borrowers with similar credit scores might behave differently and display un-predictable characteristics.

The external data used in the models do not capture the shock as a result of the current pandemic, along with the economic measures taken by governments.

Risk modelers typically resort to stress testing their models under different scenarios to evaluate the model performance. There is a definite need to carry out more elaborate testing of these models in-line with the current crisis through data simulation.

With the current risk model failing, banks are looking for other avenues to help predict customer risk more efficiently.

What banks need to do

While the nature of the current crisis is different from 2008, banks can draw similarities in terms of the failures and misses and learn from the past to plot a more pro-active path to recovery. Collection and risk managers will need to come up with intuitive solutions in short to near term as below:

Augmented Intelligence:

Post moratorium, banks will need to quickly identify customers who are at risk and in need of assistance. Collection managers are required to quickly come up with qualification criteria for these at-risk customers and treat them differently through intuitive payment plans and restructuring. There is a definite need for augmented intelligence to identify and predict such customers.

In the wake of the current crisis, banks need to be extremely proactive and empathetic to their customers. Customers will need to be respected and receive personalized treatment. Maintaining customer experience, understanding the needs, and providing optimal solutions is the need of the hour.

Post moratorium, collections managers will be required to provide a more granular and prioritized list of accounts that they need to call to their teams, as against those which can be reached through cost-effective channels such as text or emails. With a limited workforce and mounting pressure to reduce cost, collections managers will need to revamp their current operating models quickly and test new strategies faster.

Way to resiliency:

At a strategy level, collection and risk managers should start taking different measures to tackle the situation. Data analytics and advanced intelligence can help collection organizations make more informed decisions. To handle the situation, collection managers can start with the following:

  • Collect more data:
    While credit bureaus provide information about a borrower’s financial health, the personal ability to pay will be key in these challenging times. Hence it is important for banks and financial institutions to tighten the grip and have more information collected, such as borrower’s current employment details, the reason for unemployment/ income reduction, their current residential address, their family status, etc. on a more periodic basis.
  • Relook at segmentation:
    It is important to identify customers who are facing hardship and have entered collections primarily due to COVID-19 as against perennially bad customers who have a habit of missing their dues. Sub- segments will need to be created based on a borrower’s overall impact due to COVID-19.
  • Have personalized treatment plans:
    Customers facing hardship due to COVID-19 must be treated differently. Assistance programs need to be in place for different segments/ sub-segments of borrowers. Banks can look at providing forbearances or payment plans to customers even in early buckets and apply traditional collections risk strategy to non-COVID-19 related delinquent customers.
  • Reset strategies faster:
    Banks will need to be more dynamic and agile with their strategies. Strategies will have to be developed, tested, and executed as frequently as every week or earlier since these might not be valid beyond a short time frame. There is a greater need to quickly simulate strategies, implement champion- challenger on the portfolio faster, and see and evaluate results to identify what is working and what’s not. Traditional risk tools might need an overhaul with more agile tools
  • Continuously monitor their portfolios:
    While the focus is on individual borrowers, collections managers must not lose focus on their entire portfolio and look at their overall exposure to risk. Collections managers should continuously keep an eye on the impact, external factors and market changes can have on the individual portfolios. There should be a mechanism to quickly incorporate these market variables and retune the segmentation and treatment strategies.

Models need to evolve

Model risk managers will need to think out of the box and take a leaf from past disasters to develop new and more robust methods to be employed in the current scenario. Some of these include:

External factors/ alternative data:

Risk modelers have been using external macro- economic variables in the past to assess their impact on the model. However, in the current scenario, one must look at alternative data that correlates to the current pandemic and provides a view of the overall impact. Unemployment rate by industry, COVID impact on supply chain, industry sector-wise impact and revival, government policies, etc. could prove crucial in the current time. Also, identifying behavior patterns by mining insights from recent borrower call records can go a long way in determining a customer’s ability or willingness to pay.

Efficient stress testing becomes a valuable tool to test the resilience of borrowers or a specific asset class in the current scenario. Banks have traditionally been following the pre-determined DFAST and CCAR stress scenarios using variables such as GDP, unemployment rate, CPI, personal income, house/ commercial property price index, equity, market volatility, and interest rates etc. Banks need to develop scenarios that project the possible outcome of COVID-19 and suitable stress test to test their models.

A look at historical events can provide a view on the impact certain factors have on borrower delinquencies. For instance, increasing unemployment rates have a direct correlation with delinquencies. Similar past trends can be looked into for insights.

Model developers should take a more heterogeneous sample of data for their analysis and model training. In the current scenario, borrowers with the same credit scores might behave differently. Hence there is a need to incorporate more data and features at the borrower level and the economic level into the model that can help identify these subtle variances in borrower behavior.

Loans that originated close to the crisis can carry greater credit risk and will be treated differently. Separating data and analyzing based on their vintage period can provide a powerful view on the borrower’s risk.

Risk modelers also need to focus on faster time to value. Portfolio managers are in a greater need for a solution to predict and evaluate a customer’s risk on a short-term basis. Traditional timelines to build, test, and deliver models will have to be relooked at. Adaptive modeling approach that can provide robust outcomes with faster time to market is the need of the hour.

New approaches in the new normal
Correct identification of customer risk today requires an evaluation of a large number of traditional and non-traditional user attributes.

There is also a greater need to look at alternative data quickly, derive patterns, build-test-deploy- tune intuitive models, identify-simulate-test- deploy strategies, and provide a personalized treatment plan to customers with a clear focus on faster time to market.

Traditional risk models are incapable of sifting through the gigabytes of unstructured data to extract customers’ behavioral profiles.

FinXEdge by EdgeVerve offers a suite of AI-based business applications that are specifically built to solve the issues banks are facing today.

Advanced text analytics and machine learning techniques are highly effective in evaluating and quantifying qualitative aspects. AI models with the right set of data can quickly provide a granular view of customer risk against traditional models. These models can be built, tested, and tuned to help portfolio managers devise and test strategies quickly.
Swift assignment of appropriate programs to assist the borrowers in this time of need will enable lenders to build a strong and lasting bond with their customers. And isn’t that what business is all about!

NEXT ARTICLES



Accelerating Procurement Value Creation Through Intelligence

Beyond COVID-19

Like what you are reading ?

Subscribe