If banks exclude certain kinds of borrowers, this creates an opportunity for alternative lenders. This is the idea behind a host of new platforms that have emerged since the crisis.
In some cases, safe borrowers might be unable to get through the bureaucratic processes at large banks, giving rise to a kind of data arbitrage on behalf of the smaller lender. Internationally determined risk-weights or distant headquarters might override the realities of local markets for credit.
New lending platforms, especially of the fintech variety, also claim to have more efficient or innovative data analytic processes in place, allowing them to more accurately price credit.
Here’s Moody’s, today, on the topic of new underwriting technologies in general:
In recent years, rapid advances in new technologies have allowed European lenders to develop systems that enable faster credit decisions and leverage ‘alternative’ data sources, with the potential to improve underwriting outcomes. The increasing scale of data availability and storage, combined with better computational capacity has the potential to improve the accuracy of the underwriting process. New technologies, through improved predictive tools and big data access can help better assess prospective borrowers and identify those with relatively weaker profiles.
The actual list of new data "fields" Moody's mentions are as follows:
New technology allows lenders to use machine learning algorithms that could make better use of lenders' data insights, searching volumes of data that human decisions cannot assess today. Future techniques will filter for new fields like education, academic scores, labour profile, job history and other professional skills which are out of the scope of current scoring systems.
It is unclear why these data sources would help accurately assess credit risk, rather than simply restate prevailing social biases, which would increase the likelihood of mispricing. But let's suppose that it is true that "big data access" can help better assess prospective borrowers.
In an environment replete with inaccurate or poorly analysed data, you’d expect some people to be charged too much, and some people to be charged too little (intuitively, you might think lenders would skew this distribution in their own favour, depending on competitive dynamics - if there is evidence of this, we’d be interested to hear it).
So you'd expect a world with “better data” to exclude some credits that might not have been excluded before. But new data-intensive lenders do not seem to be doing much excluding. They appear to be doing the precisely opposite: lending to people who were previously excluded.
Either there are inexhaustible data arbitrage opportunities and the banking bureaucracy is completely broken (a possibility), or something is amiss.
Here’s Moody’s again, with the final clue:
The availability of new predictive systems and additional data increases the risk that lenders will use them as a justification for granting credit to financially weaker borrowers who would previously have been excluded. [...] In the US, the benefits of alternative data sources have predominantly included increasing access to credit for those with no/limited credit history.
So, we have a new kind of innovative, underscrutinised financial technology that allows credit to flow to high risk borrowers. Haven’t we heard that before, somewhere?