In this episode Jason talks to Margarita Pineda-Ucero, an ex GE Capital, Business Transformation and Risk Management expert, about crafting financial risk models during uncertain times.
Listen to this episode on Spotify, iTunes, and Stitcher. You can also catch up on the previous episodes of the Hub & Spoken podcast when you subscribe.
What are your thoughts on this topic? We’d love to hear from you; join the #HubandSpoken discussion and let us know on Twitter and LinkedIn.
If you are a CDO, or data leader setting the data strategy at your organisation, find out more about the CDO Hub.
During uncertain times, you need to be able to do the best with the information you have and be ready to pivot as necessary. You won’t get your risk-modelling 100% accurate the first time, but if you have a plan to intervene and adapt as necessary you will be a lot further along than if you wait for technology to catch up and produce ‘perfect data.’
[02:50] How does a crisis impact the ability to create financial models and risk calculations
[11.30] Why the ‘human touch’ is so important in creating models when there is limited data
[16:15] How Margarita reviews data gathering process to increase confidence in projections
[19:50] Why you need to be careful in oversimplifying when risk modelling in unpredictable times
[23:30] How to adapt to changing risk by using the first principles of risk management
Typically, when there are uncertain times ahead it is common to go back and adapt historical data to model future trends. However, since COVID-19 has immensely affected many different industries it has become too difficult to apply historical data. There have been epidemics before, but never in a modern context where technology has fostered a global online economy.
Having a pandemic in a world where there is more data available has allowed companies to extrapolate and create risk models for the future. They will not be 100% accurate but what’s of importance is, to make the most of the available information at hand.
We are now facing a new normal, and since we are not completely out of the woods yet no one really knows for sure what the fallout will be after the crisis. Take out ‘reasonable pieces of the unknown’ and look at early trends to try and create predictions. It comes with higher risk but over time you can adapt as you gather more information. If you are trying to create predictions and forecasts, be prepared to pivot quickly.
Because there is limited pandemic related data available, the ‘human touch’ has become a significant element when creating future data models. Humans can learn and adapt much faster than machines. This reactivity is extremely important in unpredictable times so that human personnel can easily intervene when needed to make necessary corrections.
By contrast, machine learning happens over a longer period of time, however the current limitations on data due to COVID-19 has only made this process longer.
Furthermore, there shouldn’t be an emphasis on whether there is more or less risk occurring. Rather, the emphasis should be on risk management and going back to the first principles of risk. For example, if you are loaning or investing money you still need to make sure that you will get your money back.
It is also important to remember that ever since the pandemic hit there has been a change in the foundation of many businesses. Customer behaviour, customer interaction and product/ service delivery have all changed. If you are able to analyse the impact of these aspects on your business into consideration and combine it with some historical data you can create a usable risk model.
Before COVID-19, risk modelers had become extremely complacent. They relied heavily on historical data and therefore only used this one main information source. The pandemic has now forced those who calculate risk to look beyond historical data and come up with alternatives to rapidly acquire and implement information.
From a risk management perspective, people that model risk now need to delve deeper into risk management and look at individual circumstances.
Before, predictions could get away with making broad generalisations about certain risk factors. Ever since the pandemic broke risk modellers have had to look at a broader range of risk factors depending on their industry. A future (extreme) example of this might be in health insurance where people might be asked whether or not someone has been vaccinated against COVID-19.
Pick out the most reasonable parts of historical data and try to combine it with the knowledge you have at hand while continuously monitoring and adjusting. Also be prepared to think about broader risk factors and how they might impact future modelling and predictions for your business.