One of the shortcomings of using historical data to predict the future is that history can’t be counted on to repeat itself. So, decision-makers have to factor in varying degrees of uncertainty into their forecasts to keep risks in check.
Recently, organizations have begun to tap real-time data to reduce this unknown element, thanks to a new set of tools that fall under the heading of Big Data technology. This trend is changing the face of predictive analytics, and allowing organizations such as Bank of Nova Scotia to lend more competitively without compromising the company’s risk exposure.
Scotiabank wanted to improve its assessment of counterparty credit exposure, or CCE, which measures the amount of money at risk if the counterparty is unable to meet its obligations. The challenge is that a counterparty’s ability to pay is often based on a myriad of unseen factors. A loan to an oil company, for example, could be affected by a drop in the price of crude, or unrest in an oil-producing country.
“[CCE] is very complicated,” says Dr. Michael Zerbs, vice-president, risk analytics at IBM, “because you need simulations; you need lots of different scenarios to make sure you really understand those risks.”
One of the most powerful features of these new tools is their ability to analyze unstructured data – information stored in text, graphic or even video format that doesn’t fit into a traditional database. This means that analysts can scour the Web looking at competitive websites, government information services, social media pages as well as data from their own applications and e-mail systems.
To make sense of this torrent of information, organizations are turning to a new breed of data scientists – individuals with advanced degrees in computer science specializing in areas such as data modelling, statistics and advanced computing.
“If you have a model that you want to utilize to predict credit scoring, in the past you may have only run it against 10 sample data sets to validate the model,” says David Smith, vice-president of marketing at software company Revolution Analytics. “Now you can run it across hundreds of data sets within a matter of minutes. That lets you have a better, more validated model for your credit scoring.”
The new capabilities have changed the way Scotiabank handles assessments. Previously, traders would ask a risk management group to run an assessment on a credit seeker, but the turnaround wasn’t fast enough to account for market changes or client requests. Risk managers, therefore, would err on the conservative side.
With access to real-time data, front-line workers are now able to make faster, more accurate assessments, allowing them to take a more aggressive stand. According to Mr. Zerbs, who helped Scotiabank implement the new software, risk measures changed by 20 per cent or more on 70 per cent of transactions.
“You can organize the way you run your business very differently if you have access to real-time information,” Mr. Zerbs says. “You can do things much more efficiently, you can empower front-line decision-makers. I think we’re at the beginning of a pretty drastic change in business models that will have a major impact that goes far beyond what we can do now.”
Analytics can also provide a company with a broader, more detailed understanding of complex relationships. Investment company American Century Investments, for example, is able to measure investments against each other, as opposed to analyzing them in isolation. “They’re not just looking at the quarterly or the daily financials of the company that they invest in,” says Mr. Smith, “but they’re actually looking at flows of money between those companies, and then using that information to optimize the performance of the fund.” Mr. Smith sees the same idea being applied to large corporations that need to optimize the allocation of resources across multiple divisions.
Analytics also offers the potential to provide solutions that are tailored to individual customers, as opposed to relying on broad demographics. Retail banks, for example, can gather and process reams of information on an individual customer and offer, say, a timely car or college loan for his or her daughter.
“Predictive analytics is letting us go back to that small mom and pop kind of experience,” says Neera Talbert, vice-president of professional services at Revolution Analytics. “We are now bringing that into the corporate world through analytics – the ability for them to understand me and my portfolio as a client.”
Risk management is essentially about hedging against the unknown. As Big Data peels back the layers to expose hidden variables, analytics, powered by real-time data, can help decode them. “It’s all about risk-aware decision making at the end of the day,” Mr. Zerbs says.
Organizations are using predictive analytics in increasingly innovative ways. Here are some areas where they are being applied.
Human resources: Because employees use their website internally, Facebook can determine how their employees interact with one another, and leverage that to build stronger teams. They even use this data source to come up with seating plans for open office areas.
Political campaigns: The Obama administration’s recent re-election campaign is well known for using analytics to target highly specific voting blocs with e-mails, pamphlets and other get-out-the-vote tools.
Product development: By analyzing social media, an organization selling individually- wrapped health food snacks targeted at women discovered that customers also wanted to buy these products for their families. As a result, they quickly formulated a packaging strategy to capitalize on this opportunity.
Marketing: Many of the targeting tactics made famous by Google are now becoming mainstream. Recently, a retail chain learned by analyzing buying patterns that a woman was expecting her first child before her husband even knew about it.