April 30, 2012 | Q&A

Accounting and Systemic Risk

Professors Trevor Harris and Doron Nissim and Executive-in-Residence Robert Herz discuss their new paper on accounting's role in reporting, creation, and reduction of systemic risk at financial institutions and across the financial system.


The onset of the financial crisis in 2008 brought greater attention to the need to understand systemic risk — the risk that weaknesses in one firm or sector of the economy will set off domino-style, successive failures in connected sectors, sending the broader economy reeling. While measures such as value at risk, or VAR, a technique used to assess the risk of loss on any given set of assets, are available to monitor risk at individual firms, such measures are limited. especially for assessing system-wide risk. And regulators and investors use their own means of interpreting firm financial reports to learn what they can about risk. But the usefulness of financial reports is moderate at best, and can add to the spiral of uncertainty driving the systemic risk at worst, when considered in the context of an individual firm and particularly when interpreted by nonspecialists depending on formulas and ratios to explain risk and a firm’s financial position — let alone across the whole financial system.

In a new paper, Trevor Harris, Doron Nissim, and Bob Herz recommend a more robust approach to assessing systemic risk, in which they use the details of JPMorgan’s financial statements as a high quality sample firm. Line by line, they show how the financial reports and accounting system aids or abets systemic risk during economic crisis, such as how infeasible it is to unearth all the counterparties including other financial institutions that are intertwined with the original firm’s business. In a special conversation with Ideas at Work, they discuss the challenges of assessing systemic risk, how accounting can and cannot help in that assessment, and what’s at stake for our financial system.

You each bring a unique perspective to this project. You, Trevor, were at Morgan Stanley until late 2008, when you returned to your academic career and when the financial crisis hit. Bob comes from PricewaterhouseCoopers and then the Financial Accounting Standards Board. Doron has a deep background in research on financial institutions and banking. Are you all of one mind?

Trevor Harris: We are not 100 percent of the same mind, but I believe that the bottom line is this: regulators can decide to measure systemic risk however they want. Investors would like to evaluate systemic risk, but given the current reporting system and information constraints, for example on counterparties, there is little chance for any substantial evaluation.

Bob Herz: With systemic risk, one has to think about different players in the whole system. Our focus is on accounting and financial reporting to investors and the capital markets. To what extent can financial statements — including disclosures as they are currently provided — give information to investors that may provide clues about systemic risk? To what extent can financial statements help managers take action that might avoid risk? To what extent does the information and accounting in external financial statements cause or exacerbate potential systemic risk?

Individual firm financial statements can't tell you about the potential for systemic risk across the financial system. Yet for the main users of the financial statements — suppliers of capital, equity investors, bondholders and other lenders, and so on — the potential impact of systemic risk on the particular institutions in which they are invested is important. So while accounting information probably can't help at a broad system level, it can potentially lead managers at individual firms to deal promptly with their exposure to systemic risk, it could inform investors about those risks, or it could potentially have negative impacts through the way people react to published financial information: what you measure matters, and what you disclose matters. We explore these issues.

TH: To be clear and to differentiate this paper from others we have written: many people look at systemic risk academically, looking at many companies and lots of data. We — mostly Doron — do some of that. But what we primarily do, uniquely, is go through a single set of financial statements, category-by-category and line-by-line to address the implications of the numbers and disclosures, to try to understand the measurement processes, and to decipher what it all means.

BH: Some people don’t get deep enough. They don't do all the “boring work” that is required to gain a proper understanding of a firm’s performance, financial condition, risks, and opportunities.

Doron Nissim: It’s important that investors and the public understand the limitations of financial statements and that there are no simple solutions. Many people believe that fair value accounting, as opposed to historical cost accounting, is useful when evaluating systemic risk. However, we believe that it is at least as important to provide consistent, comprehensive disclosures.

We think disclosures are particularly helpful because different users of the financial statements require different information. With detailed disclosures, users may be able to apply adjustments to reported numbers to make them more suitable to their specific needs. For example, equity investors are generally interested in the going-concern value of assets while creditors often focus on evaluating the assets’ “exit value.” Accordingly, when evaluating the systemic risk of an institution, creditors focus on the potential impact of systemic shocks on the current selling prices of the assets.

How should consumers of financial statements think about disclosures? What should they be looking for?

BH: I credit Trevor with borrowing from Dan Tarullo [Governor of The Federal Reserve] to help define what we mean by the key elements and causes of systemic risk. It is an important framework for purposes of this analysis:

•What are the firm’s underlying assets, how are they measured, who are the counterparties and what is their ability to pay, and what are the implications for the firm of delayed or failed payments?
•What is the liquidity of the underlying assets, especially in times of stress?
•What are the implications of price and value changes and what does this do to the solvency of the entity?
•What is the risk of contagion once a single firm experiences distress?
•And what is the relationship between the firm's business and the broader economy and financial system?

Using this framework facilitates a better analysis of where the forces of systemic risk might impinge on an organization and the extent to which a firm’s accounting and disclosures do or do not reflect the potential effects of systemic risk in a timely manner.

TH: We chose to study, after some debate, JPMorgan, because it is considered to be one of the best banks in terms of the disclosures they give. We show how different elements might be interpreted exacerbating risks and explore just how hard it is to assess systemic risk even if a firm is providing all the required data. We show just how much deeper JPMorgan’s disclosures would need to be for them to get a truly accurate assessment of the risks in their own liabilities and assets including potential offsets for those — let alone get a grip on what is happening around the system. And we are, in a sense, sending a warning signal to regulators and banks: to do this the right way, you’ve got to change the internal systems.

BH: Current financial reporting is built to suit particular objectives and purposes — to provide useful information to investors and the capital markets on the financial performance, financial postion, and cash flows of individual companies. In doing so it shows point-in-time numbers and activity for a period — but to prepare and read financial reports in relation to systemic risk you also need to understand it in a much more active and dynamic context.

TH: Doron and I have been looking at another important issue: the way income from loans is reported: it does not recognize the credit risks that are being priced in through interest costs. That is problematic and potentially exacerbates the opportunities for systemic risk.

BH: That is, it can frontload reporting of income and profits, which, when everything turns down the way it did, gets wiped clean or goes into the negative. But in the 1990s the SEC felt that to allow banks to provide in advance for potential losses would open the door to inappropriate earnings management.

TH: It hasn’t been properly understood that banks’ internal trading systems already recognize these risks as a credit valuation adjustment — the difference between the value of the asset that accounts for the risk of counterparty default and the remaining value of the asset. The better banks do this for internal reporting but are not allowed to do so for external reporting. The SEC was acting on the assumption that banks would use this adjustment for nefarious purposes, but there is a sound basis for recognizing these expected losses. The three of us have debates about the exact way to implement this, but it remains an issue in current reporting.

How big of a lift is it to improve disclosures — for both regulators and for internal accounting departments?

TH: It's big.

DN: I agree; there is much scope for improvement. But while improved disclosure will help, it may not be enough for at least some purposes. Regulators might need different information than investors, so it might be beneficial to have different accounting principles for regulatory and financial reporting purposes. In fact, this is done in the insurance industry. Unlike banks, insurance companies follow a different set of accounting standards in their regulatory reports. These accounting principles emphasize the solvency of the companies. For example, they generally use more conservative measures to value assets.

TH: It is very hard — even for internal staff — to really understand counterparty risks, no matter how good their intentions, no matter how good the systems. So they and the regulators have to look at this in terms of cross-exposures. They recognize the difficulty. One of the new regulatory bodies has put out a proposal for a rule that says every entity will have to be tagged with a standardized name so that all firms report using the same name. That will help, but it also shows just how basic some of these changes are.

Bob noted that people have to go into the detail, this is true but we should not underestimate how hard that is and how expensive it would be to implement and sustain.

And then there is the question of who is lending to whom. Greece is tiny in terms of its exposures. Why is everyone so worried about its debt crisis? Because they don't know how these layers of exposures and many credit default swaps are going to affect them. If a French bank is holding a lot of Greek debt and gets into trouble and someone has exposure to the French bank — it goes on and on.

DN: When it comes to evaluating risk, disclosure plays a particularly important role. The information in financial statements essentially reflects the realization of one possible scenario. We want information that will allow us to evaluate the company’s performance under many possible scenarios. For this, we need sensitivity metrics such as VAR or duration gaps. Fair value estimates obviously provide relevant information, but they are measured at a point-in-time and reflect current conditions; they generally do not inform on exposures. It’s true that changes in fair value give some information about ex post risk, or past risk, but this may not be very informative about ex ante risk, or future risk.

There are two important limitations of current risk-related disclosures. First, in many cases companies provide little information about the methodologies and assumptions used in calculating metrics such as sensitivity to changes in interest rates or VAR. Without adequate disclosure, financial statement users cannot determine how exactly the metrics were calculated, and the usefulness of the metrics becomes questionable. Second, because banks use different methodologies, the information is not comparable across banks, and regulators and other interested parties are unable to aggregate the information in a meaningful way to evaluate overall systemic risk.

In the paper you acknowledge that most people who use financial statements to make decisions are nontechnical users who don’t really dig in to get a deep understanding of the details. That seems problematic at best.

TH: Most analysis of business and financial markets oversimplifies fundamental characteristics of the business. Yet investors and firms use measures like earnings multiples and return on equity as proxies for understanding the value or performance of a business. These are complex institutions and businesses. As Bob and Doron said, accounting provides point-in-time measures and some flow measures but offers only limited insights. Summary statistics simply cannot provide an even closely complete picture.

In the paper you point out that there's been a focus on systemic risk in response to the mortgage crisis. But concerns about systemic risk should be extended to nonfinancial institutions as well, shouldn't they?

TH: Absolutely, there are the GEs of the world and the auto industry, which operate their own big banks. And many nonfinancial companies have finance subsidiaries that give loans to their customers. Target and Wal-Mart have them. That is one dimension. Another dimension is that the so-called real economy, or nonfinancial economy is made up in part of counterparties. So if you want to understand counterparty risk, you have to look beyond financial institutions to the underlying borrowers. The other part, speaking to Doron's earlier comment: we are a global economy. Can we know what is going on in a bank in China? Or Greece? It's an unbelievable challenge.

BH: The technology exists to do it, but you have to get constant, consistent data capture at a very detailed level, then each piece of data has to be aggregated correctly, and then you have to map that information into all these risks.

TH: Yes, the technology exists, but I can tell you from hard personal experience: firms have to spend hundreds of millions of dollars to implement this kind of system because you are fixing legacy systems that have been patched together as companies have evolved. It's very difficult to throw out what you have and start over in a new system while still conducting business day-to-day.

How have people reacted to the ideas in this paper?

TH: We've had interesting reactions to the paper. Part of the reason, I conjecture, is that we say that firms are not using the right inputs.
You can make VAR models, but the output only tells half the story because, as Doron pointed out, it is all ex post. So for four years in subprime lending, banks’ models looked wonderful. We’re saying that unless firms go deep enough and collect this data the right way — and we are acknowledging that it is really hard to do — they are not going to be able to assess risk in a meaningful way.

DN: One technical challenge that emerges when we go through the line items is that there is credit exposure and counterparty risk in each item. So, to provide measures of total exposure, banks would have to use much more comprehensive approaches than currently used. And while credit-related exposures are generally smaller for non-financial firms, aggregation issues are relevant also outside the financial sector.

BH: Clearly accounting information has limits. But it does illuminate some opportunities at both the individual institution and system levels. Right now a good deal of the reaction to the financial crisis has been to use some blunt instruments: a lot more capital, leverage caps, and other requirements. This is understandable in the absence of the right systems and technologies to gauge the extent of risks and how they change — but these blunt instruments can be costly to the whole economy. To be fair, there are important efforts underway, for example to improve the information about loans bundled in securities to provide a cradle-to-grave history of the individual loans backing a security, and there's the move to put more derivatives on exchanges. But that is only part of the solution to shaping a system that is closer to the efficient frontier in terms of both better capturing and controlling systemic risk and the overall cost to the financial system and the economy. 

TH: The Fed and the Treasury are doing what they can with what they have right now. But no one should misunderstand what’s necessary if you really want to assess systemic risk properly.

Trevor Harris is the Arthur J. Samberg Professor of Professional Practice in the Accounting Division and a senior scholar at the Jerome A. Chazen Institute of International Business at Columbia Business School.

Bob Herz is an Executive-in-Residence at Columbia Business School. Most recently, he was Chairman of the Financial Accounting Standards Board.

Doron Nissim is the Ernst & Young Professor of Accounting and Finance at Columbia Business School.

Read the Research

Trevor Harris, Robert Herz, Doron Nissim

"Accounting's Role in the Reporting, Creation, and Avoidance of Systemic Risk in Financial Institutions"

View abstract/citation 

Have You Read Columbia Business?

Columbia Business School’s alumni magazine connects alumni with each other and the School; celebrates alumni milestones and accomplishments; and chronicles the impact of Columbia Business School alumni, faculty members, and students on the global business landscape.

Read Columbia Business>