Using XBRL to Attack Systemic Risk

Already hard at work making Security and Exchange Commission filings interactive, XBRL technology now finds itself at the heart of plans to save the U.S. financial system from future calamity.

A group of risk-management leaders in the financial industry has begun studying how XBRL might bring clarity and transparency to the murky world of financial risks, much the same way Corporate America has just begun using XBRL to bring more clarity to financial statements.

While any such system is a long way off, proponents say the technology is tailor-made to help regulators (and investors) root out hidden threats to corporate balance sheets before they, well, break the bank. XBRL could, for example, let a regulator peer through a bad debt line item and see the individual loans feeding it; that task would take hours of spreadsheet diving today.

Adler

But XBRL could also do much more. Steven Adler, director of IBM Data Governance Solutions, says the computer language provides a standard vehicle for regulators to track not only weeks-old summary data, but also financial positions accruing across many banks and market segments. That would shed more light on systemic risks—which, left unchecked, can bring financial calamity of the sort we’re witnessing today.

Any potent XBRL-based scheme to report risks, however, would require the reporting of daily financial positions, a major shift in how trading firms, hedge funds, and investment banks do business. To that end, Adler’s IBM Data Governance Council is spearheading a movement that would change how investment banks and hedge funds interact with regulators.

“At this point, everybody is aware change is coming,” Adler says. “And parties would rather be in the room together talking about common solutions.”

A speech Federal Reserve Chairman Ben Bernanke delivered last month shows him to be in agreement. Bernanke advocated taking a “macro-prudential” approach to risks that are “cross-cutting,” affecting many firms and markets or concentrating in unhealthy ways. It would involve “monitoring large or rapidly increasing exposures—such as to sub-prime mortgages—across firms and markets.”

Bookstaber

Richard Bookstaber, a former risk manager for some of Wall Street’s top firms and author of Demon of Our Own Design, about the perils of financial innovation, says a markup language like XBRL is the only way to package the “massive amounts of information” involved in addressing systemic risk of the sort Bernanke discussed.

“If you’re trying to manage systemic risk, you need to know who’s highly leveraged and where there’s crowding in specific markets,” he says. “What else do these people own? The only way to know that is to get the data and have those positions. Summary risk analysis and summaries of exposure aren’t going to do it.”

“At this point, everybody is aware change is coming, and parties would rather be in the room together talking about common solutions.”

— Steven Adler,

Director,

IBM Data Governance Solutions

Bookstaber says data from the top 50 investment banks, hedge funds and broker dealers would suffice. Having such data in regulators’ hands would bring several benefits, he explains.

First, it could provide visibility across banks into “market crowding,” where (supposedly) prudent levels of risk at any one bank aggregates into extreme exposure when taken overall, risking catastrophe if the market turns.

Second, one could potentially understand how a collapse in one market might affect another. An example would be the infamous silver bubble of the 1980s, when the Hunt Brothers’ losses triggered a collapse in the cattle market because they had to liquidate their investments in this otherwise unrelated realm.

Finally, Bookstaber says, with sufficiently deep XBRL databases, regulators could conduct post-mortems on crises they failed to avert, much as aviation safety officials can pore over “black-box” flight data today to analyze an airplane crash and improve future safety.

Building on Today’s Work

Liegel

Andrew Liegel, head of product management in North America for regulatory reporting firm FRS Global, says an XBRL taxonomy of risks could also help financial institutions improve their own internal risk and reporting analytics. He calls the disparity of risk formats today “disturbing.”

XBRL AND RISK

The following excerpt is from the blog of Steve Adler from the IBM Data Governance Council.

The XBRL Risk Taxonomy work we’ve begun has the potential to reshape the dynamics of financial regulation. But XBRL is just a tool, not a solution. The solution that an XBRL Risk Taxonomy can enable is standardized loss reporting from financial institutions to regulatory authorities and back again.

What do I mean by this?

Every regulated financial institution would provide loss event and tail data to regulatory authorities in XBRL via the Risk Taxonomy. With thousands of institutions reporting losses on a regular basis, the repository would grow large quickly, but meaningful trend data would still take at minimum a few years to accumulate. But over time, this repository would not only provide regulatory authorities with a risk pulse on the financial system, but would also enable financial institutions to compare their own loss reporting to industry aggregates to improve trending and forecasting. The key is to link this loss trending information to management decision cycles so that every decision point can be compared to past experiences and future forecasts. This is not to make already risk averse executives hide beneath their desks, but rather to enlighten human decision-making with risk probability information at the point of action, record decisions and results, and constantly learn from mistakes and improve over time.

We humans do this on an intuitive level every day, but our best decisions are dependent on human memory and painful reminders of past individual failures. To combat systemic risk resulting from incremental action, human experience needs to be captured, profiled, and broadcast to more humans who may have an interest to regard what others disregard. This creates autonomic opportunities as well as governance checks and balances.

The Insurance Standards Organization performs this kind of loss data aggregation for a variety of insurance lines and many insurance companies purchase this data via subscriptions to calculate insurance premiums and reserves. Sadly, despite the rapid rise of insurance-like hedging strategies on Wall Street, like credit default swaps and portfolio insurance, no one is using traditional insurance products to cover Operational exposures and without loss history insurance carriers can’t price coverage or buy re-insurance.

With several years of loss data accumulated it should be possible to create an open insurance exchange to underwrite the losses with insurance coverage. This would allow banks to transfer operational risks (which are most similar to professional liability exposures) off their balance sheets to insurance vehicles. The banks would pay a premium for the coverage, the market would price risk on a near-real-time basis, and regulators like the SEC could govern premiums and fair trade mechanisms. In some ways this would function like credit default swaps but the trades would be on an open market, and rising risk in financial institutions would result in higher premiums, which in turn could be correlated with equity and bond markets to create additional incentive and penalty mechanisms for risk management.

I think this idea has enormous benefits for many market participants. Risk self-insurance is inherently inefficient capital allocation without deep loss history. In the insurance market self-insurance is most practical when commercial coverage is unavailable or too expensive. In the banking world, banks have self-insured their own losses for decades without empirical risk measurement programs.

Today of course the Taxpayer is providing catastrophic insurance coverage for banking failures and that is the most inefficient coverage in the world!

A better model would price future risks based on past losses and make banks pay premiums for loss producing behavior. The XBRL risk taxonomy can create a data model to facilitate loss history aggregation that can create enough data for accurate underwriting. And when that information can be placed on an open market, banks would have a financial incentive to report losses—because the market would transfer the losses to insurance coverage and banks would have more capital for investments. At a time of capital constraints, this solution has something for everyone—market mechanisms, regulatory reform, and better capital allocation.

What’s hard about pricing operational risk coverage is the long tail of losses. Traditional insurance policies, with fixed duration, deductible, incident and aggregate policy coverage won’t scale to the volume of loss events and severity tail. An exchange, however, can price large volumes of loss events and tail growth in near-real-time, providing both incentives and penalties for poor risk management in firms that transfer via the exchange. That in turn will transform loss reporting from the cat and mouse game it is within firms today into a business necessity because every unreported loss is a balance-sheet deduction in capital allocation that will get penalized severely by the market when reported late.

Turning this solution into reality will require a new Risk Information Management infrastructure in financial institutions, regulatory authorities, and market exchange mechanisms. It will depend on a common data model, standard risk measurement reporting processes and technologies, and cultural changes on Wall Street. This is why we are starting with an XBRL Risk Taxonomy to standardize loss reporting.

We can’t create solutions like these overnight, but by starting with common reporting standards we can inspire a 21st Century infrastructure that regulators can build upon to enable risk analysis and oversight at nearly the same speed the market participants create it.

Source

Steve Adler of IBM Data Governance Council Blog (Feb. 2, 2009).

People in finance and accounting often discuss risk in a different language than regulatory compliance groups, he says. For example, a bank’s credit department might weigh risks based on the number of loans with payments 30, 60, or 90 days overdue; the compliance group might assess the same information using probability-of-default bands.

Atkin

XBRL is one way to capture and convey that data across groups, says Michael Atkin, managing director of the EDM Council, a non-profit working on a “semantics repository” standardizing the terms and definitions of financial industry data.

But XBRL itself will only be “the raw materials to look at risk in lots of different ways,” he says. Defining the terms for a dictionary of financial risks will be the big challenge.

Some types of securities—government and corporate bonds, foreign exchange, equities, commodities—should be straightforward “barcoding,” Bookstaber says. Derivatives and swaps present a bigger challenge, given their tendency to “defy classification because they combine the properties of insurance, taxation, indexes, and bonds,” as Adler puts it. What’s more, these securities often trade off-exchange and without contracts.

But derivatives are ultimately based on some asset, and therefore should be definable, Bookstaber says. Such definitions will require cooperation among regulators, technology firms, and (above all) financial institutions themselves. And “cooperation” may be a euphemism for “government mandate” to get the banks to part with their definitions and data, Adler and others say.

Turner

“There’s range of information that banks and hedge funds and other investment actors have been completely unwilling to part with,” says John Turner, CEO of CoreFiling, which develops XBRL software for financial reporting. Even once a taxonomy does take shape, Turner expects hurdles in gathering the data.

“We’re dealing with very sophisticated systems inside organizations that built them themselves,” Turner says. “There are also questions about how this data would be protected; it would be terribly, terribly sensitive.”

Turner says such an XBRL-based financial reporting system would have to be international in scope, with regulators collaborating using common information across Japan, Europe, and the United States. He put the number of participating financial intuitions at 500.

Jon Wisnieski, a senior information systems specialist with the Federal Deposit Insurance Corp., says regulators’ access to fresh data on financial positions would be a big step forward. Today, he says, the FDIC data on derivatives can be 45 days old.

“Really what we’re talking about is transaction type of information being passed on to a regulator, which would be very valuable,” he says. Wisnieski says the FDIC’s experience establishing its own XBRL filing system, the Central Data Repository online since 2005, could be instructive.

“Smart people can figure out the technical issues. It’s more the soft issues, personalities, backgrounds, and culture that you need to get over,” he says.