Risk Managers have never been short of data – they are usually drowning in it – so they can be forgiven for not getting too excited about the prospect of ‘Big Data’: the idea of orders of magnitude more digital information becoming available about everything.
Risk managers monitor the variation in performance of their companies – financial, operational, human, ethical – to keep it within acceptable limits. So the first impact of Big Data is that there will be a lot more things they can monitor in much more detail: activities, interactions, signals, movement of goods, financial fluctuations. The need to sample subsets of information or take snapshots of partial views will be replaced by having data on all of it, all of the time. So the biggest transformation will be the need for interpretation. Risk managers will adopt new techniques for visualisation of vast clouds of data, hunting for the needle in the haystack, spotting the anomalies, looking for outliers, and identifying weak signals of abnormality. They will look at connections between things, view the world as networks, detect time progressions and evolving patterns, and learn to construct and read data maps the way that early explorers invented cartography.
Even more importantly, they will change the way they analyse and interpret, using much more powerful computational resources than we are used to today. Those of us who grew up optimizing analytics to make the best use of scarce computing resources may be surprised by the way future generations use it when it is abundant and free. Risk management is about envisioning and contingency planning for possible future problems. Perhaps the risk managers of the future will generate billions of stochastic possible future projections. They may compare thousands of different opinions, or use every different scientific interpretation to interpret model uncertainty.
Risk and uncertainty go together. Today analytics is all about getting good data to reduce uncertainty: “garbage in, garbage out”. Data theorists are now telling us that lots more data, even if it is of much lower quality, can reduce uncertainty on estimation. Forget trying to clean up your modelling data, just find more indicators of similar things and let your algorithms do the heavy lifting.
So what might it actually mean for risk management and the business decisions that Big Data might support? Could it enable better pre-emptive action to minimise the chances of a weak anomaly escalating into a crisis? Might it shift the balance from crisis management post-event to preventative management before it happens? Could it give managers greater confidence to detect signal from noise, or to understand the cost-benefits of making decisions probabilistically? Could it change the very nature of risk management itself?
The dynamics of risk management is set to change significantly. Change often means sacrifice and surrendering things we currently value. Big Data threatens to be intrusive and compromise our privacy. Thinking through the opportunities that might offset this cost is a vital part of sharing expertise between specialists in venues such as the Cambridge conference. Big Data could have big benefits for reducing society’s risk.