by Roman Hagelstein
A generation ago, analytics software was a rarity at banks and insurance brokers. Employees mostly had guidelines and crude balance sheet indicators for denying or approving credit to customers. We’ve seen a silent arms race since.
Today, data analysis reigns supreme in determining which factor predicts if a client will pay back its loan. The big data models in use in every bank’s credit department are constantly optimized. Only niche players can survive without hundreds of interconnected Stata workstations, constantly fed by information on types of customers and their default rates. The computers outperform us humans for auto loans or mortgages, because with the right guidance they can reveal patterns in the company’s customer base that would overwhelm humans alone.
But what do banks do if they are faced with a question, but don’t have enough data? They are, of course, forced to rely on human judgment. The logical next step, then, is to improve in this area. The insights that banks can gather from crowd intelligence projects are rather straightforward.
The first step is to train employees in making better predictions. That is easy enough! They simply need to use base rates for an initial forecast, balance inside and outside information, not relying too much on either. Then they should adjust whenever relevant news becomes available. They must remain actively open minded throughout, still looking for information that goes against their own views. The harder part is to keep practicing, learning from the inevitable mistakes as well as the successes. The path to being a Superforecaster requires a lot of work and keeping score along the way so you can get the feedback that makes you better.
Will businesses, in the future, fully automate their business strategy? I think not. Relying on superforecasts is not automation. It may feel like putting the company on auto-pilot. But the enhanced judgment calls will still have to come from people. In addition, early adopters of a more scientific approach to forecasting will only have a temporary, initial advantage. My prediction is that all large banks will be using enhanced forecasting of business strategies within twenty years, with a likelihood exceeding 85%.
Roman Hagelstein, from Frankfurt am Main, Germany, has been a Superforecaster with since 2014. He works in Financial Controlling near Frankfurt, and has also worked in a variety of corporate finance and audit roles after he was awarded a Ph.D. degree in Economics in 2010.