Blog
Michael Turner
Credit Data Gaps Exist, But They Can be Filled
Consumer and Micro-, Small, and Medium Enterprise (MSME) lending has increasingly become driven by data. This data-driven lending has increased credit access and has resulted in fairer, more objective lending. However, needless to say, data driven lending requires data. Those who do not have data collected on them to properly assess credit risk may not only not benefit from easier access to credit but may find themselves increasingly shut out from lenders that use highly automated lending approaches. In the context of MSME lending, the challenge stems from a lack of data on both individuals and businesses given that, for smaller MSMEs, access to credit is typically a function the proprietor’s credit worthiness. So, what impacts the owner’s credit access, can impact MSME access to finance. Fortunately, there is a solution to the credit data gap problem, adding new sources of data to the traditional credit record.
We are increasingly aware that payments for non-financial services, such as energy and water utilities, pre-paid data from mobile network operators, and even rent can populate credit reports. Research by PERC and others has shown that this non-financial payment data can be highly predictive.[1] Add to this other types of predictive data such as address stability, assets, education, public records[2], and unstructured data such as social media (Facebook, LinkedIn, Twitter, Yelp reviews)[3], and we see on the horizon the potential for closing credit access gaps.[4]
While some of these new approaches may be better suited to more developed economies with very robust data environments it is likely that emerging markets will benefit the most from utilizing new data sources in lending. In many emerging markets that have very low rates of traditional credit bureau coverage (such as 5.8% in Sub-Saharan Africa)[5], coverage by non-financial services, such as mobile subscriptions, can extend to the majority of the adult population. And for some developed and developing markets, public data offers another potential means of extending financing. In developed economies, income and employment data are often collected by governments to administer unemployment insurance and could be made available for underwriting purposes. Publicly held data in some emerging economies also show some promise. In Tanzania, newly required electronic fiscal devices collect and transmit transaction data—including VAT identification number, item sold, quantity sold, unit price, and date of the transaction—from businesses for tax monitoring purposes. This data can be used to develop instantly generated basic accounting statements for lenders.
Data from growing online market places also offer promise. In the US, Kabbage underwrites business loans by accessing real time business transactions data from a variety of sources such as eBay, Amazon, business checking, and PayPal.[6] In China, Alibaba’s Sesame Credit Management seeks expand access to credit among small companies. Using data such as online shopping and sales, e-pay of utility and telecom services and address stability, Sesame Credit is moving to fill substantial data gaps.[7] Despite the existence of these data sources, existing systems of collecting non-financial data remain few and far between in emerging economies. They will have to be built. Many data sources exist that can be integrated in systems that can speed underwriting and scoring. For MSMEs, supply chain and other transactional data can now be recorded with relative ease via handheld devices.
Yet, we should note that the reasons for large, consequential data gaps in lending have more and more to do with market and policy failures, stemming more from incentive structures, than from technical limitations of gathering data. The business model of traditional credit bureaus (not paying for data, for instance) may be misaligned to acquire non-financial data. Misaligned incentives exacerbate a data sharing challenge that is already often significant owing to issues such as the lack of a national ID and poorly constructed privacy/data protection laws.
To address these barriers PERC is working on a project called Financial Identity Risk Management or “FIRM” that would compensate non-financial data furnishers, utilize biometrics where available and needed, and not store data in a central source but access it in a pass-through model with an applicant’s consent. So, FIRM would not aggregate data but in a hub-and-spoke fashion access it and make it available to lenders when needed.
FIRM would form a symbiotic relationship with traditional credit bureaus in a given market, and would access data not already collected, or likely to be collected by credit bureaus. Then, when a lender extends credit based on a FIRM report, it will report repayment of the traditional credit to a bureau. That is, FIRM’s success is also beneficial for the traditional credit reporting system.
The world is awash in data. The same IT revolution that enabled automated underwriting and credit scoring is also creating vast amounts of data and new solutions. Data gaps need not be a major problem in lending.
[1] See http://www.perc.net/publications/research-consensus-confirms-benefits-of-alternative-data/
[2] See Lexis-Nexis’s effort: http://www.perc.net/publications/research-consensus-confirms-benefits-of-alternative-data/
[5]See http://www.doingbusiness.org/data/exploretopics/getting-credit
[6]https://www.kabbage.com/how-it-works/
[7]http://www.ft.com/intl/cms/s/0/34e77fe8-a6a3-11e4-9bd3-00144feab7de.html#axzz3a2J4p31V
