Mark Davison, Chief Data Officer, Callcredit Information Group

Missing a lucrative opportunity can cost you dearly. Just ask the ‘almost investors’ of Amazon. Back in the mid-90s, Jeff Bezos, the founder of the retail behemoth, approached family, friends and others for a $50,000 investment in his idea for an online book shop. Those that could who believed in the idea invested. Those that couldn’t, according to Bezos, to this day find it painful to talk about anything Amazon related.The lesson to be learnt? Missed opportunities could haunt you for years and, in some cases, for businesses, could lead to their demise.

Currently, a big opportunity for businesses is machine learning, which is making waves and transforming industries, leading to significant improvements in service and productivity. But it’s not commonplace yet in credit risk and financial services, despite the fact that its benefits are varied – including helping businesses to comply with regulation. It’s time to ask why, and prepare for a future in which machine learning is at the core of financial decision-making.

Industry scepticism

A stumbling block in the move towards machine learning in the credit risk sector is the scepticism about the technology that’s pervasive amongst risk professionals – and needs to be overcome as a first step. Although its benefits are clear, in order for businesses to make it work effectively, they need a significant depth of data and a critical understanding of the problem domain they’re working in.

However, fears about the ability to explain the output from the models has led to financial services businesses being hesitant to adopt machine learning until they are confident it won’t cause any issues.For example, risk analysts for creditors may be reluctant to let a machine make decisions instead of people (or pre-existing trusted algorithms). And fraud prevention specialists may have more trust in traditional systems to tell them an application might be fraudulent. On the other hand, their counterparts in other industries,such as retail or customer service, are testing new technology much more boldly and moving ahead at pace.

To overcome the scepticism around machine learning, we must recognise that it already surrounds us, even if we aren’t always aware of it. It drives the internet and every interaction on it.

The time is now

But why has machine learning grown in popularity now? The use of artificial intelligence (AI) and machine learning technology began some time ago but has really been gathering pace over the last few years. What’s different now is the availability of quality data and the power of computing we have available, as well as more understanding from regulators around the world of the effectiveness of these techniques. It’s now easier than ever to use machine learning to make more informed decisions on lending or credit. We have reached the tipping point when businesses need to stop thinking about it and start actually using machine learning.

Falling behind

Yet, credit risk professionals are already playing catch up when it comes to the use of machine learning. Fintechs, established hedge funds and banking firms alike have already introduced it to their technology portfolios, and early adopters will be the ones with the competitive advantage.

Although it requires a significant leap for risk and credit professionals to go from not accepting machine learning as an option to it becoming an integral part of credit risk processes, it’s crucial that they begin the journey now.

Better decisions

Once risk professionals catch up with other sectors, one of the key benefits of machine learning is that it will enable them to make better decisions – especially as regulators start to relax the constraints that lenders and fiscal institutions work within. This will include encouraging greater competition in the market, demanding greater transparency around decisions and practices from those who are already there, and driving more intelligent use of data such as social media wherever it is available.

As the credit risk sector places a greater focus on transparency, machine learning will be used more frequently by lenders as a new way to solve existing problems. Risk and affordability modelling necessitates a constant explanation of lending decisions, ensuring that the data going in is in line with the eventual decision.  Using a model called supervised learning, machine learning will allow lenders to analyse large quantities of data. In this process, it will discover relationships that until now have been hard or impossible to see, before making a decision. This will allow lenders to use the machine’s output to justify its decision to a regulator or a consumer.

Backing up the claims

So how do we know that it works? At Callcredit, we ran a year-long machine learning trial to help showcase the potential benefits of the predictive accuracy that machine learning can offer. The study showed encouraging results and points to potential financial benefits for adopters of the technology in the credit, fraud and insurance sectors.

In one modelled scenario, the level of default in a portfolio of 60,000 credit cards was reduced significantly, resulting in a 10 per cent decrease in overall bad debt. If used with other elements of the customer lifecycle, potential benefits generated as a result of machine learning could be even greater.

Machine learning is not going to get robots to do human jobs or change the nature of what creditors do. Instead, it will get the same tasks done more efficiently and in a more intelligent way, complementing systems that are already in place, and helping experts to improve the lending process.