Predictive modeling has an image problem -- at least where financiers are concerned.
At the recent CFO Risk Management Summit in Boston, panelist Minakshi Srivastava, vice president of U.S. card acquisitions at Bank of America, said she always gets a good amount of pushback from bank colleagues when presenting advanced analytics data. "They say, 'It's good, but can you predict the future with it?'" James Costa, panelist and chief risk officer at TCF Financial Corporation in Wayzata, Minnesota, seconded that. He said because predictive modeling never saw the financial crisis in 2008 coming, "it was almost deemed part of the problem."
But, as economists have noted, the models used then were not up to the task at hand. Economic models assumed a certain level of stability around trends and variability, in line with "historical averages," according to a recent article from The Economist. That's not a great assumption for detecting a crisis. Even so, while the economic meltdown may have soured some financiers on predictive modeling, it hasn't turned them off completely. "I don't know how you handle risk, which is almost by definition the science of uncertainty, without models," Costa said.
Since the crisis, the federal government has mandated financial institutions undergo a formal stress test at least once a year. Used to anticipate how the bank might perform under unfavorable conditions such as high unemployment rates or a spike in oil prices (and, let's hope, toxic mortgages), stress testing has evolved into a, "prove that you have the capital" test, Costa said. "It's one more thing you need to do to demonstrate to board and regulators that you've done your homework and, yes, can take on more risk."
The stress-testing mandates are driving a couple of trends for financial institutions. It's increasing collaboration. "Very disparate groups, at least in financial services, are required to work together," Costa said. It's also increasing a demand for model risk management. Experienced personnel who understand how financial models work -- and can take into account their inherent risks -- are hard to come by, Costa said. And here's where the CIO's ears should perk up: Stress testing isn't new to IT, which uses a similar practice to test how systems will respond to dire circumstances such as hack attempts or distributed denial of service attacks. That's expertise any finance department can leverage -- regardless of the industry.
iOS 8 could be a big deal for CIOs
iOS 8 is scheduled to be released for general availability Sept. 17, which could turn out to be a big deal for the enterprise.
"For the first time, there's going to be app-to-app communication capability," said J Schwan, CEO and founder of Solstice Mobile, a mobile solutions provider in Chicago. An application will essentially be able to share and pass information to another, which in some cases will eliminate the need for manual entry or copy/ paste commands for passwords and even credit card numbers, he said.
And in other cases, it will create new experiences altogether. Apple's new health app HealthKit, for example, will integrate exercise, nutrition and even sleep data from applications like RunKeeper and wearable devices like Fitbit into a single location. Apple is also partnering with healthcare giants like Epic Systems, the electronic health record software company. Users will be able to share information between HealthKit and Epic's MyChart app -- not in the cloud, but on the actual mobile device.
"That data is transferred on the phone," Schwan said. "And the individual person has control of whether or not that information gets shared." Some want to characterize this kind of data integration capability as big data, but Schwan sees it differently. "We call that a small data transaction."
The new app-to-app capability will be a boon for consumers, but it could also transform how the business does, well, business. "From the CIO or enterprise perspective, it gets interesting because now my phone can be my hub," Schwan said. The line of business will be able to, say, move a customer's data from a CRM app into a document management system app. "Instead of all that integration having to happen in the data center, that integration can happen on the phone," he said.
As CIOs lead mobile app development, Schwan recommends redefining business processes to fit these new capabilities for external and internal customers alike.
When should a chief data officer report to a CIO?
Gartner Inc. predicts that by 2015, 25% of large global organizations will have appointed chief data officers. One question frequently asked about the role is who the CDO should report to. Talk to enough experts, and you'll hear everything from the CEO to the CIO to the CMO.
Glenn Finch, global leader for technology and data at IBM Global Business Services, said companies right now are pretty evenly split as to whether the role sits inside or outside of technology. In his view, the decision should hinge on the analytical sophistication of the business.
If the business is "analytically voracious," constantly asking IT to fix analytical data, that's a situation where a CDO could comfortably sit inside of IT, according to Finch. But, if the CIO is inspired to push for a CDO because everyone else is doing it, "then I'm going to go be a solution looking for a problem -- a hammer looking for a nail," he said.
"The [greater the] culture shift that has to happen in the business to adopt an analytical, decision-making framework, the harder that role is to sit in technology," said Finch, whose team recently released a new report about this emerging title. "What the CIO doesn't want to do is force the role to be established, have it report to him or her and then struggle for the business to identify value."
It's also important to note that, because the position is still so new, the organizational structure that works today may not be what works tomorrow. "As the position matures, and as organizations mature, you're going to see that role move around a lot," Finch said.
Previously on The Data Mill
Struggling to open up data? Look to the public sector
IT and the business at odds with defining big data success
Consumer messaging apps make their mark on the enterprise