By Michael Baker, Managing Director, Scope Risk Solutions GmbH, part of the Scope Group
What for you are the benefits of attending a conference like ‘Risk EMEA’ and what can attendees expect to learn from your session?
One of the biggest benefits is the opportunity for “cross-fertilisation”, by which I mean learning from subject matter experts in different areas of credit risk, such as consumer lending, structured finance or large corporates, and whose organisations have different risk appetites and exposures. Personally, I am focussed on lending to large and mid-sized corporates.
Whilst I don’t know what delegates don’t know, I hope they will obtain insights on the benefits of technology already being captured, those on the horizon and those which might be “pipe dreams”.
In your opinion, how is technology changing approaches to credit risk?
The biggest impact of the advances in computing power has been the ability to assemble so-called “Big Data” and then to analyse and interrogate it efficiently. Automated decision-making has been of benefit to consumer lenders in particular. The impact on corporate credit analysis has been less pronounced, but we are seeing an improved ability to supplement fundamental analysis of financial statements with more forward-looking indicators, derived from both public and private data. Examples are greater use of key word searches and analysis of payment behaviour. The roll-out of xBRL in many geographies has also been positive
Have AI and machine learning had much of an impact on credit risk? What key challenges have arisen since their introduction?
The initial impact has been on the asset classes where the borrowers are more homogeneous – consumer finance and SME lending. The impact on corporate lending will be on the identification and analysis of “Alternative Data” – information that provides leading indicators of a firm’s performance well in advance of the publication of the financial statements. Examples are using Machine Learning to power “key word” searches of social media, using technology to predict economic activity via air pollution (and thus macro-economic trends) or better analysing payment behaviour.
The challenges are manifold. Certain players are in the prime position to exploit big data – Google, Amazon, eBay and large supermarkets, rather than traditional banks. The regulatory environment is struggling to keep up with the pace of change. As Professor Fei-Fei Li of Stanford University puts it: “There’s nothing artificial about AI…It’s inspired by people, it’s created by people, and—most importantly—it impacts people. It is a powerful tool we are only just beginning to understand, and that is a profound responsibility.” Regulators and traditional lenders are faced with a skills shortage, exacerbated by organisational cultures which are quite different from the firms in the forefront of AI.
Could the application of artificial intelligence help new modelling techniques? What would be the key benefits of this?
The continued increase of computing power may well help develop new modelling techniques, although in my opinion the techniques to date have evolved rather than been revolutionised – certainly it is now easier to identify clusters in data and generally recognise patterns. At present we are in the midst of a crisis and no doubt slow-moving credit risk models will not have predicted the defaults observed.
If indeed we move to a “new normal” historic data may no longer be as relevant and building new datasets which are both deep and wide will be most helpful. That said, until this year, models in common usage were performing to quite a high degree of accuracy, so the upside in terms of both unwise lending avoided, good loans not advanced and the associated pricing might be limited.
What do you see ahead for technology advances within credit risk?
I think Natural Language Processing (NLP) will improve and contribute more to corporate analysis, both in terms of examining Reports & Accounts and in creating and analysing alternative data. It will facilitate analysis of social media, websites and search engines. That said, firms might modify their behaviour in the light of such changes. Certain players (e.g. supply chain finance specialists or credit insurers) are in possession of particular elements of data and should be able to leverage that further. As lenders gain more timely access to management accounts via Cloud Accounting, Open Banking and possibly Enterprise Resource Planning (ERP) and CRM systems they can use AI to formulate more forward-looking indicators to supplement traditional fundamental analysis.