The information that banks have on their customers can be a treasure trove of data. But, making full use of it has been difficult due to the enormous volume. One shop that is doing something about it is Citigroup. They have been using the IBM “Watson” computer (of “Jeopardy” fame) to process millions of documents, including unstructured data such as e-mails, news reports, books and websites. (Eventually, there are hopes of expanding the capability to materials that only human beings can analyze now, such as newspaper financial pages and company reports. This information could be used for predicting takeover targets, and for making smarter investment decisions.
Some areas where banks can gain from big data analysis now include fraud detection, analyzing credit worthiness, and complying with rules on money laundering and sanctions checking. Checking client names against a sanctions blacklist can be a particularly tricky process, since bank customers with the same names are common. It can be both embarrassing and financially costly to wrongly deny a legitimate customer bank services, due to an identity mix-up.
To deal with this dilemma, massive amounts of data from many different sources are needed to detect enough distinguishing characteristics (such as nationality, address, transactions with other countries, etc.) between two similarly named customers to correctly identify who is who.
The problem of detecting fraudulent transactions can become more acute as money transfers come from a variety of sources now, such as computers and mobile phones, and payments shifting from cash to cards and electronic transfers. It adds to the difficulty when transactions involve multiple countries. Finding the often obscure patterns that can raise a red flag require sophisticated algorithms making thousands of checks across millions of transactions, all in time to accept or reject a funds transfer. Big data analysis makes this possible.
The massive amounts of stored customer data can also allow a bank to take far more than the 15-20 factors that would traditionally be taken into account to determine the precise interest rate to offer on a mortgage or other loan. More exciting for the banks is the potentially opportunity use this sophisticated data analysis to individually customize approaches for selling customers additional bank products.
The point is that the changes taking in big data acquisition, number crunching, analysis, and ever more inventive uses of this information will fundamentally change the ways that banks do business in the future. There will be many issues to still to deal with. For one, banks’ data storage capacities are already falling short of what is needed to house the data that is being created, and it is expected that that gap will only to widen. For another, how will the smaller banks compete, when the investment to gain this advantage may be cost prohibitive for some of them?
But these challenges will be dealt with and sorted out, and these trends will only grow.
For a much more detailed look at this topic, you can refer to the primary source of this material, the May 19, 2012, issue of The Economist magazine’s Special Report on International Banking.
