[From NexBillion blog, 13 January 2015, with David Porteous]
In Spent: Looking for Change, the recent documentary about financial exclusion in the United States (embedded below), there is a moving segment about a young man named Justin who is determinedly rebuilding his life after having obliterated his credit rating by failing to repay his credit card debt. He says (from the 16:00 mark): “People often judge me on the choices I've made, not knowing the options that I had.” Maybe if we knew the limited options Justin had when he decided not to repay his debt, we would agree with him that he had taken the most appropriate, even responsible, action by not repaying. If that were the case, wouldn´t that make us want to trust him more rather than less? Years later, when his situation and options had changed, we would likely feel positive about offering him a new loan for a new beginning.
Economists say that credit bureaus are about solving information asymmetries between creditors and borrowers. But which asymmetries are the economists talking about, exactly? No credit bureau helps Justin explain to financial institutions that he was forced to scratch out a living entirely on his own from age 16, that his earnings didn´t always last to the end of the month, but that those days are now behind him. All the credit bureaus do is to propagate information on his past non-repayment.
As David Graeber argues compellingly in his sweeping history of money Debt: The First 5000 Years, we now take for granted that all loans must be repaid, fully and on time, as if that was a natural societal imperative - but that has not traditionally been the case. Debt moratoria, renegotiations, substitution for tokens of assets or labor, even wholesale cancellation of debts, are recurring themes everywhere. It is only fairly recently that debt repayments have become an absolute test of character, often summarized in a three digit score. Credit scoring has become something about you, disconnected from your circumstances and options.
With the digitization of finance, we face the daunting prospect of “the system” having an unforgiving and unforgetting memory of poor people´s formal debt repayments while knowing little else of any real significance about them. Credit providers will build a profile of you based on disjointed circumstantial evidence, slowly and painfully crossing datasets, almost accidentally – you have $20 in your savings account, a mobile ARPU of $1.70 per month, demonstrated successful repayment of a $10 instant mobile loan – but you can blow whatever positive attributes you’ve demonstrated all on one unpaid debt. One strike, you´re out. Big Data can become the basis for a new exclusion.
Exclusion is often the result of ignorance. Ignorance creates prejudice: in the absence of concrete information, you generalize. And big data is fundamentally about drawing big generalizations - excuse me, correlations. I can easily believe that these correlations will work increasingly well on average, giving new financial opportunities to many. But in the process, many Justins will be pushed further into financial untouchability, silent casualties of the unfathomable wisdom of the machine. Something that demonstrably works on average may be commercially satisfactory, even socially impactful, but may fall short on inclusiveness – in the sense of helping those who need help the most.
A way to prevent these unfortunate side effects is for financial service providers to avoid single prescriptions and actively work towards enabling multiple avenues for credit, so that people like Justin can have several shots at getting the credit they need. Providers should recognize that they might never get close enough to poor customers living in marginalized communities to fully understand their evolving circumstances and options.
So in addition to seeking ways of collecting more data on prospective borrowers with which to make individual credit decisions, lenders should also seek ways to build more trusted relationships with local players in each community, who they can count on to make credit recommendations and help channel credit within their network of friends, customers, savings circles, and business associates. Justin´s credit need not come directly from a bank, it could come from a local store that understands the true nature of his troubles, or a neighbor who knows how he´s turned himself around.
In a recent paper, we develop this theme by exploring three different Pathways to Smarter Financial Inclusion: by serving poor people directly, reaching poor people indirectly through the businesses within their communities where they work and buy things, or using social networks as (informal) distribution channels. Note that all three pathways rely heavily on clever analytics, but the objective of each is subtly different: the first draws credit inferences from the little data that is specifically available on poor customers, the second creates sufficient business intelligence and data flow to be able to underwrite local traders´ and entrepreneurs´ credit decisions, and the third creates incentive structures for peer screening and monitoring.
Finance will reduce people´s sense of vulnerability only if it creates a hierarchy of options in people´s minds. It´s not necessarily the case that people want more credit, they just want more options. By moving beyond traditional methods of collecting and interpreting borrower data, financial service providers can help provide those options to people who are often excluded.