In the coming weeks the feds and the surviving financial services institutions will have the daunting task of unraveling all the securitized loans and other instruments that are hiding the toxic investments. But does the technology exist to do that? And if so, could it have been used to prevent the bad debt from hitting the fan in the first place?
The fact is that despite government regulations like Sarbanes-Oxley, there is little visibility mandated by current regulations into the origination of loans and how they are broken up, resold, and resold again.
To cite the classic example of how we got into this mess, consumers were given 100-percent-plus variable mortgages without any security. Not only could those mortgages be sold to other banks, but they could be divided into five, ten, or twenty tranches -- financialese for slices -- and resold to five to ten different organizations, making it difficult to track who was involved and who ended up taking the risk.
Theoretically, the financial service providers were clear on the risks of each type of loan and had a way to gauge whether they had enough liquidity -- cash and other easily sold assets -- available if the riskier loans went south. But a New York Times report indicates that in fact many financial institutions gamed their analytics to favor positive scenarios over negative ones in order to justify keeping less money in reserves should the risky loan blow up. "A large number of buyers of these kinds of instruments really didn't care about the value. They just wanted to flip it. A lot of people just didn't want to know," concurs says Josh Greenbaum, principal at Enterprise Applications Consulting.
Analytics and CEP Tools Could Have Helped
Had these financial services companies and banks established business intelligence metrics as to the ratios of what kind of debt they were holding versus the cash reserves they held, their analytics systems might have driven alerts earlier in the process, says Michael Corcoran, a product manager at the BI provider Information Builders. But as anyone in business already knows, consolidating that kind of data to get those answers more often than not is a slow process that typically ends up being done manually in an Excel spreadsheet well after the fact.
Jeff Wooton, vice president of product strategy at Aleri, a complex event processing (CEP) company, agrees that most data consolidation takes far too long to give a complete picture. "It relies on overnight data consolidation runs, overnight reports, and manual processes like spreadsheets."
That's where technologies such as CEP and operational BI come into play. They analyze huge volumes of transactions -- 100,000 messages per second with millisecond response time -- and can set up alerts and even trigger remedial actions by other systems. Such tools, such as Aleri's Liquidity Management System, already exist to help treasurers in global banks gauge their liquidity position in real time. Wooton says that over the last two weeks there has been considerably more interest in such products than in the past.
Wooton cautions that the various kinds of analytics tools available, such as business activity monitoring, decision support software, data integration, and alerts, could have offered a warning but not fixed the underlying problem of financial services firms misjudging -- and in some cases, misrepresenting -- the risks of their loans and securities.
But taking analytics, CEP, and data integration to the next level to give regulators a sense of what all the financial institutions were doing and what the liquidity risks actually were could have helped, and could prevent a recurrence, Corcoran says. He says that the use of middleware could bring the data together and create a "common front end" that is shared by regulators and the services companies alike.
But that front end needs a common back end, especially around the data that should exist, says consultant Greenbaum. "The data model for doing the analysis doesn't exist," he says, so a company selling securities and packaged mortgages doesn't include the packaging history. "You can't do the classic drill-down," Greenbaum says, because no one knows what the relationships are because the metadata hasn't been preserved -- or at least not preserved in a way that is easy to find.