The fiscal crisis of 2007–2009 caught people—including expert economists—off guard. And even as the crisis unfolded, useful insight was hard to come by. "At the end of the day, most policymakers and economists had some idea as to what was going on, but not as much as one would hope," says Arvind Krishnamurthy, a professor of finance at the Kellogg School of Management.
Why should this be? Why were the over-leveraged, so-called "toxic" real estate assets that contaminated the rest of the economy invisible even in plain sight? According to Krishnamurthy, the problem is that financial reporting practices are too outdated to capture the information necessary to accurately assess modern macroeconomic health.
"In trying to document the crisis [after the fact], you start to look for the data you need and you find it's not exactly there," Krishnamurthy says. "Our measurement of financial activities might have been good in the 1940s or '50s, but it's very poor currently given how much the financial sector has changed."
Current Financial Reporting Falls Short
Current financial reporting focuses on assets and liabilities on a cash balance sheet. "If you look at an industrial firm like GM, their assets are the various manufacturing plants. These are all tangible things, and when you look at the financial statements, you can see it all fairly clearly," Krishnamurthy explains. The trouble is that firms like Goldman Sachs and Morgan Stanley assume contingent economic exposures, through financial instruments like derivatives, that do not show up clearly on their balance sheets.
"A simple derivative security might be one in which, when real estate prices go up by 10%, this company will make an extra $5; and if they go down, they'll lose $5, so the risk is commensurate to the $5," Krishnamurthy continues. "But you can’t really see the $5 of risk on financial statements. You see that the firm has a derivative, perhaps tied to real estate, but you can’t see if the firm is exposed to real estate risk of order $5, $50, or $500. Right away this obfuscates things."
Risk, rather than cash instruments, says Krishnamurthy, is the attribute of our financial sector that defines its overall health. And this risk goes largely unreported in financial statements to the government or the public. "If you work at Goldman Sachs, the way you measure your own risks is entirely different than these types of [public] accounting statements," he says. "You'd measure things like, 'How much money do I make or lose if real estate prices go up or down by 10%?' That's the principal number you're interested in."
The other attribute that must be measured and reported more accurately, says Krishnamurthy, is liquidity. "The classic example of a liquidity problem is a bank that has a bunch of assets in real estate loans that are all coming due over the next 20 years. On the other hand, its liabilities are short-term deposits from account holders," he says. "The bank can potentially get themselves in a liquidity problem because the assets are long term and far less liquid than its liabilities: If all the depositors come to the bank and ask for their money, the bank will have to close up because the money's coming in over the next 20 years. The modern financial sector is perpetually in this position of 'short liquidity,' and it's important to measure how much they are short liquidity because liquidity meltdowns are central to financial crises."
The irony is that while few financial firms report this data, "most companies collect it for internal purposes in fairly sophisticated ways," Krishnamurthy says. Unlike assets and liabilities on a balance sheet, risk is fundamentally dynamic, so assessing it requires a different approach from classical financial reporting. "You can't get at it if you take a single snapshot—you have to ask what-if questions," he continues. "This is what most financial firms do internally. They construct what their financial positions will do in given scenarios, and then they aggregate."
"The truth is that it's easier to not invest in standardizing data reporting and remain relatively opaque. But this is where a government can step in for the common good." — Arvind Krishnamurthy
So-called "stress testing" is another, more public, method of measuring risk and liquidity. "A bank constructs a what-if scenario: say, what if unemployment goes up and stock market falls by this much, how much money will the bank lose?" Krishnamurthy explains. "That's what the Federal Reserve did in 2009, and now it happens every six months for all the big banks."
Externalizing and Aggregating Risk Assessments
However, these ad-hoc, often internal risk assessments cannot help paint a broader picture of the economy unless they can be externalized and aggregated. Few firms are motivated to do this, since it is expensive and the letter of the law does not require it. Also, firms may have different "technological languages," as Krishnamurthy puts it, for describing their risk and liquidity. These languages would have to be standardized across the financial sector in order for useful reporting to occur.
"The truth is that it's easier to not invest in standardizing data reporting and remain relatively opaque. But this is where a government can step in for the common good," Krishnamurthy says. "One way to do this is expand regulatory filings that require banks to use a common language. It seems possible that if everyone communicates risk in the same way, eventually they will disclose it on, say, SEC filings."
What this kind of new financial reporting could enable, argues Krishnamurthy, is a "50,000-foot view of the economy's 'risk map,'" which would illuminate pockets of toxic risk and illiquidity in the overall landscape, like those that caused the recent financial crisis. Krishnamurthy's paper "Risk Topography," coauthored with Markus Brunnermeier of Princeton University and Gary Gorton of Yale University, takes its title from this analogy to mapping.
Krishnamurthy compares the measurement of risk to that of the gross domestic product (GDP) during the Great Depression. "It was obvious right after the Depression that measuring GDP was relevant, but nobody was doing it in the early 1900s—the data didn't exist," he explains. "We are in a similar situation today. We'd like to have a risk map of the economy drawn, say, every six months as the terrain changes. Our hope is that, in 15 or 20 years, we'll have accumulated data that some bright PhD student will use to design a brilliant theory about the causes of financial crises. But the student will need the data about risk to build it with. That's where we're starting."