Regulating Financial Guarantors: Abstraction Bias as a Cause of Excessive Risk-taking

Since the 2008-2009 financial crisis, scholars, regulators, and policymakers have engaged in extensive studies to try to control excessive risk-taking by systemically important financial firms. In a recent article, available here, I argue that those studies do not fully explain the unusually excessive risks taken by financial firms as insurers of bonds and other debt securities, as sellers of protection under derivatives known as credit-default swaps (“CDS”), as providers of credit enhancement in securitization transactions, as issuers of standby letters of credit, and otherwise as guarantors of financial obligations (collectively, “financial guarantors”). Despite their sophistication, financial guarantors tend to underprice risk relative to other financial firms. They also fail at a much higher rate than do other financial firms.

We need to better understand financial guarantor risk-taking because of the immense size and impact of the financial guarantee industry. For example, the notional value of the CDS market alone has been as high as $57.8 trillion and remains in the tens of trillions. Bond insurers had been insuring hundreds of billions of dollars of debt securities, and the industry may be poised for a revival with the potential to aid in the development of infrastructure projects worldwide. The standby letter of credit market likely is in the trillions, with such instruments being used in a wide range of transactions, including lease agreements, stock purchases, financial security, commercial paper, and trade investments. There always will be a need for financial guarantees; institutional investors such as insurance companies and pension funds often are required to invest a large portion of their assets in AAA-rated securities. Because most securities have lower ratings, these investors need financial guarantors to enhance the credit to AAA levels. Furthermore, the influence of the financial guarantee industry is greatly multiplied by the systemic impact not only of a guarantor’s default, but also of its rating downgrade, on its counterparties.

My article hypothesizes that financial guarantor risk-taking is influenced by a cognitive bias in perceiving risk, which I call “abstraction bias.” The business of financial guarantors focuses on committing to pay out capital only if certain future contingencies occur (such contingent pay-out commitments hereinafter referred to as “financial guarantees”). In contrast, the business of virtually all other financial firms focuses on paying out capital at the outset of a project, such as a bank that makes loans or a financial institution that makes investments. Because financial guarantors do not actually transfer their property at the time they make a guarantee, they may view their risk-taking more abstractly (the abstraction bias), causing them to underestimate the risk even after discounting for the fact that payment on a guarantee is a contingent obligation.

Recognizing abstraction bias as a cause of guarantor failure can inform financial regulation. It can help, for example, to refocus regulation away from facile solutions – such as the Dodd-Frank Act’s requiring securitizer risk-retention, or “skin in the game,” while ignoring that the market has always required such risk-retention – to solutions that address the more complex realities.

Abstraction bias also helps to solve a puzzle that several scholars, including the author, argue has been distorting post-financial-crisis regulation. Politicians and the media have attributed much of the crisis-related excessive risk-taking to the so-called originate-to-distribute model (OTD) of securitization, in which originators of risky loans sell them to third parties. These sales are presumed to transfer risk on the loans away from the originators, thereby creating moral hazard that encourages originators to make even riskier loans. That explanation fails to explain, however, why those third parties – or why investors in and financial guarantors of securitization transactions sponsored by those third parties – accept that risk. Although risk marginalization can help to explain why investors individually might accept that risk, it cannot explain why a financial guarantor – which takes on most if not all of a transaction’s risk – might do so. Abstraction bias helps to explain that failure.

Some might question whether abstraction bias is related to the endowment effect – the cognitive bias that we value what we own more than what we do not own. It is, however, fundamentally different: The endowment effect does not apply to commodified assets, such as money. Nonetheless, abstraction bias and the endowment effect appear to have a common root in loss aversion – the observation that people weigh losses more heavily than gains in evaluating potential risks and outcomes.

After examining abstraction bias as a hypothesis, the article examines evidence of its existence. As mentioned, financial guarantors fail at a much higher rate than other financial firms, suggesting they engage in unusually excessive risk-taking. Consider, for example, the case of bond insurers. Of the nine firms operating prior to the financial crisis, all but one failed – a failure rate far higher than that of banks. The AIG debacle provides additional evidence of unusually excessive risk-taking. An AIG subsidiary sold so much protection under CDS contracts prior to the financial crisis that its parent, one of the world’s largest and most sophisticated financial institutions, was forced to post massive amounts of collateral to the swap counterparties. Ultimately, the federal government bailed out AIG, fearing its failure would have greatly exacerbated the crisis’ severity.

The article also empirically, and more systematically, tests abstraction bias by comparing the pricing of otherwise parallel risk-taking by financial guarantors and by financial firms that pay out capital at the outset of a project. The comparison shows that financial guarantors significantly underprice risk relative to those other financial firms. That helps to confirm that abstraction bias is real. It also confirms that abstraction bias can influence the behavior of even sophisticated financial guarantors.

Finally, my article analyzes how the law can be used to try to correct abstraction bias, adapting recent studies that show how cognitive biases can be addressed and sometimes improved. This includes applying the approach that professors Jolls and Sunstein call “debiasing through law.” Recognizing that even the best regulatory efforts are unlikely fully to correct abstraction bias, the article also examines how to try to mitigate the harm that that abstraction bias can cause.

This post comes to us from Steven L. Schwarcz, the Stanley A. Star Professor of Law & Business at Duke University School of Law. It is based on his recent article, “Regulating Financial Guarantors: Abstraction Bias as a Cause of Excessive Risk-taking” available here.

1 Comment

  1. RW

    Does your article factor in the Revenue Recognition rules of the SEC and the accounting industry? Maybe I’m wrong, but as I understand, in the years before the 2008 crash AIG was allowed to report, as current income, the long-term earnings estimated to come from the many CDS protection contracts it entered into. This boosted AIG’s net income and, not coincidentally, the bonuses paid to AIG officers based on that income (and share price). I have not heard of any claw-back of those bonuses.

Comments are closed.