Africa’s economy is under pressure. In South Africa, the finance minister recently downgraded 2017 economic growth forecasts to 0.7% from 1.3%. In Kenya, political uncertainty has prompted the Treasury to warn that it, too, may have to revise economic growth projections downwards. And Nigeria only just recorded its first period of growth after five consecutive quarters of retraction.
In tough economic times, one of the first things consumers default on is their credit repayments. With the festive season approaching – and with it, the temptation to spend more on credit – consumers are slipping further into a debt trap that they can’t get out of – and banks are taking the brunt of it.
Banks are facing increasing pressure from local and international regulators to better manage their credit books and to properly vet customers before granting them loans. It’s as much for their own protection as it is for the consumers’. The only way to do this is by applying advanced analytics to internal and external customer data so that banks can build statistical models that automatically screen customers for credit-worthiness. However, uptake of these types of solutions in parts of Africa has been lagging.
Banks have until January 2018 to comply with new International Financial Reporting Standards (IFRS 9), which require them to not only predict and make provisions for future credit losses but also to provide a full audit trail of how they came to those calculations. That means they have less than two months to get the data, models, infrastructure and skills in place to avoid the reputational impact of being non-compliant with IFRS 9, which would reflect non-transparency in financial reporting, as well as the bank’s ability to do business. Credit scoring is important in determining some of the models and parameters used in meeting IFRS 9 requirements.
But African banks face two major hurdles to compliance. First, they often don’t have access to the volumes of data needed to build effective credit scoring models and, if they do, the data is often “unclean” and scattered across disparate databases that are not integrated.
Second, some banks have not invested in the tools and skills needed to build these models. This forces them to outsource model creation, which has its own disadvantages, including high costs, late response to market changes, and ending up with models that are not aligned to their portfolios.
Yet, credit scoring is now imperative for all banks – and their ability to gauge whether a customer will repay their loans every month or if they are likely to default, will determine whether banks stay on the right side of the regulations.
There are three main elements to credit scoring: customers, banks and regulators.
Ideally, consumers should be taking measures to strengthen their credit records – by paying their accounts on time, spending within their means, and limiting their use of credit cards. But the reality is that, when credit is available, it’s difficult to resist using it.
The upside is that banks have access to this information about customers – how much they owe, what their spending patterns are, how much they earn, etc. This information forms the basis for credit risk scoring and is invaluable to banks when it comes to processing loan applications or for identifying new customers and opportunities to upsell to existing customers.
The data is available but preparing it and ensuring it is clean and correct is the biggest and most time-consuming part of building predictive data models. In fact, there are reports that some data scientists and modellers can spend up to 80 percent of their time just getting their customer data in order.
Banks face a balancing act of either issuing a high number of loans – and facing the risk of customers defaulting – or issuing fewer loans but compromising their ability to generate revenue. Now, new regulations are forcing them to balance their risk and return profiles to ensure that they are profitable while controlling risk within their loan portfolios.
To throw a spanner in the works, competition in the market is getting fierce, with new entrants competing for a shrinking pool of quality customers.
Using advanced data analytics and credit risk scoring models, banks get a single, accurate view of their customers. By continually training and updating the models, banks will be alerted to problem customers as well as to opportunities to sell new products to existing customers with high credit scores. This gives them a competitive advantage while allowing them to manage their risk exposure.
Model monitoring is crucial to ensure they are performing accurately and are using the most recent data. Basing decisions on outdated and inaccurate models could expose banks to risk by issuing loans to customers who would not normally qualify. This is one risk of outsourcing model development rather than developing them in-house. By the time the bank implements those models, they could already be out of date.
In an attempt to avoid a repeat of the 2008 financial crisis, financial regulators are implementing stricter compliance measures. Apart from the new IFRS 9 accounting standard, banks also need to start preparing for Basel IV – just when they’ve gotten to grips with Basel III. Basel IV revises methodologies for the determination of capital requirements, which means that capital calculations across all risk types will change.
In the case of both requirements, regulators are becoming more stringent on governance and control issues around methodologies and data that banks used to develop the credit scoring models and to determine how much capital they need to cover future losses.
Outsource vs DIY
While under certain circumstances it is appropriate to buy ready-made generic credit models from outside vendors or to have credit models developed by outside consultants for a specific purpose, there’s a case to be made for bringing data modelling in-house.
Today’s advanced data analytics solutions address a number of challenges facing African banks: they’re easy to use and allow anyone to experiment with data visualisations and drag-and-drop interfaces; they pull data from disparate sources into one interface – and prepare the data for modelling; and they ensure that models are retrained based on real-time information, reducing risk exposure for banks. Banks will also benefit from economies of scale when many segment-specific models need to be built. Further, building a solid, internal skills base also makes it easier for banks to remain consistent in the interpretation of model results and reports and to use a consistent modelling methodology across the whole range of customer-related scores.
Until now, African banks have been slow to adopt advanced analytics solutions for credit scoring but with increasingly stringent regulation, they no longer have a choice. The added bonus is that analytics can increase their competitiveness and reduce their risk exposure – without massive investments in skills, time or infrastructure.
By Charles Nyamuzinga, Risk Consultant at SAS