October 24, 2018
Effective risk assessment is at the core of the lending business. Retail lenders have their own risk assessment models based on traditional frameworks. Meanwhile, technology companies are developing risk assessment models increasingly inclusive of new variables derived from data generated as a result on increasing internet and smartphone usage. These models provide an opportunity for lenders to expand their target audience to the previously unbankable population.
Sound risk assessment requires adequate market knowledge, deep understanding of the customer, and robust systems to detect fraud and forgery.
Risk assessment is difficult due to multi-fold challenges, the most important of which are:
Low coverage of credit bureaus: Just 43.5% of the Indian adults are covered by credit bureaus. The lack of a bureau record forces the lenders to reject the applications of more than 40% of the borrowers.
High operating cost: Information about the majority of the Indian customers is unavailable and unverifiable digitally. Lenders are forced to perform verifications manually – field verification, phone verification, reference checks, etc., increasing the operating cost of risk assessment. Some estimates suggest that traditional lenders’ have an operating cost 3X higher than that of digital lenders.
Highly informal economy: About 45% of the Indian GDP and 82% of the Indian workforce are associated with the informal economy, making it difficult for lenders to source trustworthy data for verifying identity, documents, addresses, etc.
The challenges can be easily mitigated if the lenders have access to data sources providing detailed and accurate information about borrower’s profile, behavior, and preferences.
Alternative data consists of information that enables comprehensive risk assessment and is not a part of traditional credit bureau reports. Retail lenders can deploy alternative data technology for better customer service delivery, capturing new segments, improving cost efficiency, and enhancing the underwriting process.
With an almost 450% rise in the number of smartphone users the last three years and with internet penetration increasing by 10 million every month, there has been a steep rise in customer-generated digital footprint in India, which in itself is a rich source of alternative data.
Alternative data can be obtained from multiple sources, where the customer leaves behind a digital footprint. All these data sources provide valuable insights about the borrower. For example, online spending data provides insights about the borrower’s profile; a borrower spending substantially on online betting may not be considered ideal by lenders.
Each transaction, SMS, and email contributes to the pool of information and provides micro-insights into one’s risk profile. The micro-insights enable lenders to develop a more comprehensive customer profile for underwriting, which is being consumed by retail lenders across the globe to enhance their risk assessment.
Source: International Finance Corporation (IFC)
For the graph above, the higher the area under the bowed curve, the more efficient the risk-assessment model in segmenting between a regularly paying customer and a delinquent borrower.
IFC suggests that incorporating more data in the credit risk assessment models will push the model curve upwards, increasing the area between the bowed model curve and the 45-degree line and hence will improve the model’s ability to distinguish between a good borrower and a bad borrower.
Alternative data provides insight into the customer and these insights can transform into metrics or scores which can be consumed by retail lenders for risk assessment.
Retail lending risk assessment consists of three major modules:
Lenders have traditionally relied on income estimates and the debt-to-income ratio as a metric to determine the ability to pay. Surrogates generated from alternative data enable lenders to complete a holistic assessment, more so in primarily informal emerging economies such as India. For example, a sum of monthly cash deposits into an account in addition to payments received through cards or m-wallets such as Paytm is a substitute for income of a mom-and-pop store owner in a tier-2 Indian city.
Credit history and performance on previous loans have been traditionally used to assess the borrower’s intention to pay.
With a sizable demand for credit coming from new-to-credit borrowers, it is growing at 16 million every year. With a very low credit bureau penetration, alternative data sources provide an economic opportunity for creditworthiness assessment. Telco data, rent, and utility bills payments history act as a substitute to assess intention-to-pay for a new-to-credit salaried borrower.
Operationally costly and manually intensive, lenders go through field visits, multiple calls, and manual verifications to ascertain basic personal information like identity, address, etc.
Verification costs can be reduced with the adoption of innovative technologies and solutions. With over 90% Aadhaar penetration, the number of Indian social media users increasing at 31% YoY and availability of accurate location data due to a 500% increase in the number of 4G-enabled smartphones from 47 million in 2015 to 218 million in 2017, lenders can verify the identity and addresses as well as identify the home address and work address for a mom-and-pop shop owner in a tier-2 Indian city.
Incorporating alternative data marks a landmark change in the complete risk management process for retail lenders.
Like for any mom-and-pop shop in a tier-2 city, lenders find it difficult to assess credit risk for most of the 64 million micro, small & medium enterprises’ owners, and millions of salaried Indians who are not covered by credit bureaus. Alternative data provides the lenders with additional sources to gather insights about the borrowers.
Alternative data has the potential to improve risk assessment for retail lenders. To capitalize on the alternative data sources, retail lenders will have developed cutting-edge data science and technological capabilities.
Traditional lenders have consistently partnered with technology companies and digital lenders to enable themselves with capabilities to extract value from alternative data.
For example, the Foundation for International Community Assistance (FINCA) partnered with First Access for what they claimed to be the largest alternative credit scoring approach by a microfinance institution (MFI) in the world.
In 2016, India’s largest bank, State Bank of India (SBI) launched SBI e-Smart SME, a working capital loan offering for sellers on e-commerce platforms, which moves away from traditional lending based on financial statements like balance sheet and income tax returns; and instead uses proprietary platform data and surrogate information from public domain to assess the seller’s creditworthiness for loan sanctioning. The product is launched in partnership with Snapdeal – India’s largest online marketplace, with the widest assortment of 20 million+ products across 500+ diverse categories from thousands of regional, national, and international brands and retailers.
The power of alternative data is believed to be revolutionary. It remains to be seen whether the lenders will give it a chance or not.