You shall not pass
On risk and risk management in fintech
For financial industry risk function is often a stepchild. It is seen as growth inhibitor and time-eater, so presumably without it any company could be moving forward much faster. However, while sales department makes the top-line skyrocket, risk takes care of the (future) bottom-line and helps financing providers to move into black-ink zone, while avoiding the snake pit of fraud. It is relevant for fintechs as well, who can outrun their bigger counterparts using technology-driven approach to smart risk management.
Brave digital world
It is very much true that technology creates easier, cheaper and faster processes for the financial sector, unveiling mechanisms once too complicated for the end users to grasp. Digitisation, being a general development vector for the whole industry, eliminates human error, erases borders, disregards time zones and moves cash flows at an enormous speed. Risk management benefits from it as well.
For lending in particular going digital means increased transparency of credit decisions, which are based on previously insignificant criteria. Technology also provides a larger set of potential investors and borrowers with access to lending market. For B2C companies great emphasis is placed on customer satisfaction, and sole traders are literally inseparable from their businesses. So what can be more obvious than taking into consideration Facebook rating of a corner shop, buyer evaluations of an Amazon trader or Yelp reviews of a local family business? Enters the technology, helping the lenders instantly pick bits and pieces of information about the borrower, varying from external credit scores to website statistics. This leads to better informed lending decision and flexible client performance monitoring.
Risk modelling has also been rapidly evolving in the last years. While traditional financial institutions are able to accumulate loads of information simply because of their long history, fintechs have the advantage to look over the fence of common metrics in their modelling. As for the data sets, those could easily be acquired for analytical and training purposes. Broader view of the underlying patterns makes tech-based companies more flexible when it comes to improving or correcting the models.
Technology also does wonders to fraud identification and prevention. Increased number of data points, possibility to reveal not previously obvious patterns and identify irregularities in customer behaviour, to name a few, help immensely.
Bug in the ointment
Well, everything looks just perfect, so are there any risks left? As my job is to see the glass half-empty than half-full, they are a few. Leaving data protection and cyber security beyond the scope of this article, digitisation gives birth to more creative ways of its misuse, be it outright fraud or unconscious error.
As said, fintech uses statistical modelling extensively and leverages available (big) data. Skilfully built and calibrated risk models discriminate between decent and no-go customers with better accuracy, taking into consideration trends and factors once neglected by more conservative constructs.
Still, eagerly passing final credit decisions to models can cause one’s default rate growing almost at the same speed as hockey-stick revenues. Lack of proper accounting, low financial and technical literacy, young company age, diverse languages and currencies, certainly add to the mess of banking the underbanked. Self-reported information is quite often misinterpreted, deliberately or not, which could cause chaos when being fed into the model. What is more, streamlined lending that relies fully on automation, becomes a target for fraudsters, who use imperfections of unsupervised process.
All this could be seen as inevitable trade-off when moving to all-digital from “gut feeling” but the reality is not that gloomy.
Humanising touch
Let’s say, as long as there is a human being on a borrower’s side, one needs to balance them with a human on a lender’s. But fear not, risk departments in fintech still consist of people. That does not necessarily mean a bunch of underwriters painstakingly slowly processing every single credit application. The human analysts are there to adequately react on each case outside of model scope… and make sure that next time a similar one appears it is already been reflected in more advanced approach to risk evaluation.
Incredibly important thing humans do is model re-assessment and calibration on a regular basis. Financial and thus lending landscape has changed drastically in the last decade. Quantitative models working under pre-2008 assumptions when mortgage crisis hit would quickly drive banks and other financial companies to collapse, should they have neglected to review their risk appetites (i.e. put temporarily on hold financing of construction companies).
The same stands for more recent boom of tech startups and the following crisis. Profitability, previously considered as one of the main metrics in company evaluation, does not make much sense in case of growth-oriented tech firms. Once they are in the market to search for cheaper money than venture capital can provide, they end up in a grey zone for traditional financial companies. That is when fintechs can take over, feeling easy in a pool of information which could drive big banks crazy.
Long story short, digitisation of financial industry certainly means progress as well as inevitable “exchange of one nuisance for another nuisance”. While technology is an important tool to stay on a competitive edge and deliver faster, higher and stronger, do not forget about human beings on both sides of the process. The most recent Nobel Prize in economics celebrated long-standing contribution to behavioural science for a reason. There is nothing wrong for the clients to be irrational, after all, it is just only right for fintechs to manage it accordingly.
