How environmental factors influence loan performance
There has been a lot of noise regarding artificial intelligence in risk management, but the overlapping chatter is often confusing.
To understand if and how this technology can help financial institutions manage their credit risk, we have to first understand how an ocean of data can grant us the power to foretell the fate of a loan.
And so, we’re going for a stroll down the hall of computational biology.
I can hear you already. “Biology and finance – what on earth could they have in common?”
We’ll get there. But let’s start on a light note… cancer.
Infamous for its quality of being unpredictable and deadly, it remains one of mankind’s greatest foes.
But how do you know if a cancer patient will beat his illness? Can you tell whether he will improve or worsen? Survive or perish?
No, you can’t. But machines can.
Predicting the course of a disease
After great leaps forward in technology, we have been able to apply deep learning paradigms to cancer research in order to foresee the unforeseeable.
When studying cancer patients, the organisation has to decide whether the patient is recoverable and has the possibility to recuperate in order to decide the next steps.
By studying the mortality curve and taking into account certain factors, we can predict the outcome or course of a disease and are able to:
1. Define future therapy, or
2. Accompany the patients with treatments such as chemotherapy until their last moments
To determine this, the organisation observes multiple variables.
But which ones? Is it a patient’s behaviour? A patient’s gene signatures? Or is it his environment that influences his ability to recover from the illness?
For an accurate prediction, we must take into account all variables.
For example, both the gene signature and the environment influence a patient’s ability to recover greatly.
If you put a sick patient in a healthy, sterile environment, he is more likely to survive. Put him in a toxic environment and his chances of recuperating grow slim.
And at this point, we can draw the parallelism between computational biology and finance from a mathematical perspective. How?
Well, the bank’s approach to loans is quite similar.
The genetics of a loan
When a loan is in arrears, the bank must decide what action to take.
To do this, it examines the personal “financial signatures” to understand the person’s propensity to risky spending behaviour, then it decides whether or not to commit resources to help the borrower “recover”.
However, like with healthy organisms in toxic environments, this borrower may show promising gene signatures while living in a context that might increase his chances of defaulting.
Likewise, a not-so-promising individual may happen to live in an environment that will contribute to his economic growth.
There is no doubt that a loan originated in a risk location will influence the probability of default.
And yet, many financial institutions use a very limited amount of traditional data to determine risk, despite there being an abundance – and even and excess – of data.
That means sticking to data from a credit bureau, a credit application or a lender’s own files on an existing customer.
This data is useful – but does it give you the whole picture?
There is a wall between you and your customer. And this kind of data represents a small hole through which you hope to catch glimpses of your customer’s life.
But numerous other things are happening on the other side of this wall, things that you cannot see and, therefore, cannot take into consideration.
This can easily lead to errors in judgement which, in turn, would put the financial institution at risk and might also result in a high number of missed opportunities.
That being said, you can’t jump over the wall to see what’s going on.
At least, not entirely.
The data providers of today homogenise data from multiple datasets across the world, which provides the bank with the macroeconomic frame of where the loan resides.
With this, we return to our previous discussion on healthy organisms, or loans, placed in toxic environments. Understand the person and the environment and you will more accurately predict how the loan will perform. Of course, for now, such data will not give you a 360° view of your customer because there are always additional variables to be considered.
However, if you want to start chipping away at the wall that stands between you and the customer, this is the place to start.
Once that wall starts to crumble, you will be able to see your customer’s situation from different perspectives instead of from that one tiny glimpse.
But this is all easier said than done without the appropriate technology and data.
Luckily, there are plenty of solutions out there that can help you take advantage of the growing amount of data in our world.
In fact, this is a recurring topic across all industries, not just financial services.
If you would like to know more about how to access and use the right data in a credit risk setting, please feel free to reach out to us.