The financial world is not without risks. As part of their fiduciary duties, banks operate with intense systemic risks every day, while empowering small firms to large multinationals begin new ventures, innovate and incubate, and ultimately act as the custodians of trust. If not carefully monitored, these systemic risks can easily snowball, and this can impact not only the banking network, but also the financial health of the country at a macroeconomic level.
A real-world scenario
When a corporate takes a loan has taken loan from a bank for building a plant. Normally, the bank will disburse the loan amount in tranches, using which the borrower will continue to pay suppliers for plant construction. Now, let us assume that the borrower attempts to defraud the bank by diverting the loan funds. In the guise of making vendor payment, the borrower sends the money to a “distributor” – a shell corporation that exists only on paper. Can the bank be defrauded?
Let us examine the whole gamut of information to which the bank is already privy:
- The list of approved parties with whom the borrower is expected to transact for the project. Source: Project documentation
- The fact that the sum transferred to the “distributor”, is very similar to the amount disbursed by the bank for the project. Source: core banking and transaction systems
- The purpose of availing each loan installment. Source: the CA certificate to be mandatorily submitted to the bank
- Whether the distributor is a blacklisted company or in the news for the wrong reasons, subject to frequent tax raids or audits. Source: Big data
If all of the above information (available at the different branches and locations of the same bank) can be shared with the authorities on time, the bank can proactively prevent fraud and save itself from an unsavory and litigious situation involving painful asset quality deterioration.
Credit risk management in perspective of the RBI mandate
The premise for Early Warning Systems is set here It all begins with credit risk. Broadly speaking, there are two aspects to defaults – ‘inability to repay’ and ‘no willingness to repay’. Both could potentially result in NPAs or Non-Performing Assets.
Following the Asset Quality Review of 2015, the RBI rolled out a string of regulations mandating the adoption of EWS as a best practice in identifying and mitigating the risks posed by Red Flagged Accounts (RFAs). The guideline issued mandated systems that would consider 45 indicators of stress in borrower accounts, measure the accounts with respect to each indicator, and flag incidents to authorities. Indeed, a laudable effort from the RBI.
Technology service providers were able to unearth much more data on borrowers from big data sources in addition to traditional data sources and this has aided with insightful decision-making:
- Massive data ingestion and analysis of loan portfolios of banks across the country, products, and industries/customer segments, to take management calls on pulling back or expanding credit to specific sets of customers
- Detect the stress of borrowers from what is reported in semi-public sources, including legal cases, share pledging, dubious business dealings, and so on.
- Listen to rating agencies on what they are saying about their borrowers, industries or the economy
- Listen to online and social media chatter on the promoters of a borrower
Now is the opportunity to leapfrog from just analyzing the transactional data of borrowers’ accounts to looking at them strategically.
To begin with, the sanity of data itself is a big factor. The key is in knowing where to look for data and when, and this is no easy task. If we examine industry-leading banking risk management systems, like rt360 built by BCT Digital, which are custom-built for the Indian banking sector, they have some of the most extensive sources and credible touchpoints, making data compilation all the more effective. The advanced algorithms and rules engine are extremely effective in mitigating false positives and unwarranted alerts – an area that is particularly hard to manage. The rules as such are far more exhaustive; for instance, rt360 is configured to flag 200 warning scenarios – well more than the 45 proposed by regulators.
How Artificial Intelligence is transforming Early Warning Systems
Early Warning Systems rely on tens of thousands of data points to measure and monitor risks, which is almost impossible for humans to replicate. Artificial Intelligence can transform Early Warning Systems, enabling them to make instantaneous predictions and extract actionable insights from disparate data sources, using these four distinct transformative components:
- Collating data from multiple touch points
- Cleansing, validating and restructuring data into valuable information
- Algorithmic processing using next-generation technologies and data modeling to generate insightful early warning signals/alerts
- Case management by channeling alerts to decision-making authorities
The future will reveal to us the role of EWS in strengthening asset quality. Furthermore, for the system to achieve its full potential, there needs to be open collaboration between the bank and its technology partner, and this is where partnering with a service provider that has specialized risk management expertise is bound to show results.
The struggle that banks face in combating credit risks is multi-dimensional. Foremost among them is determining the right capital allocation and pricing for different sources of their revenue.Read More