SAFE Finance Blog
02 Mar 2022

What the financial crisis teaches us about bank disclosure and loss recognition

Financial stability depends not only on the timeliness of banks’ accounting information during the crisis, but also on the capital regulation setting incentives for prompt corrective actions when losses are first recognized

The 2008 financial crisis set off a major debate about the role of accounting for financial stability. Early in the crisis, some policymakers, regulators, and academics were quick to criticize fair-value or mark-to-market accounting (FVA)—the practice of recognizing assets based on market prices, arguing it exacerbated the crisis by facilitating excessive leverage in booms and leading to contagion and downward spirals in busts. With respect to the role of FVA, a substantial body of evidence has emerged since the crisis and this work provides many important insights (see for example Ryan 2008, Laux and Leuz 2009, 2010, Badertscher et al. 2012). Collectively, this evidence suggests FVA did not contribute to or exacerbate the financial crisis. The allegations against FVA were largely based on hypothetical links and models, rather than empirical facts.

In our paper, “Accounting for Financial Stability: Bank Disclosure and Loss Recognition in the Financial Crisis”, we re-direct the focus on those accounting issues that have received less attention during and after the crisis, yet are central when it comes to the link between accounting, reporting and financial stability. These issues are the lack of reliable risk disclosures, the delayed recognition of loan losses, as well as the countervailing effects of prudential filters on managerial incentives to take prompt corrective actions early during the crisis.

Fair-value accounting did not play any major role during the crisis

We begin with a brief summary of the evidence on the role of FVA that has emerged since the financial crisis:

  1. FVA, let alone pure mark-to-market accounting, played a very minor role for most banks. It is a misconception that FVA dominates banks’ balance sheets (e.g., Laux and Leuz 2010; Becker et al. 2021). For banks, loans constitute by far the largest category, and a large fraction of banks’ losses occurred in the loan books. Banks use amortized historical cost accounting for almost their entire loan portfolio, even when they have the option to use FVA.
  2. When banks applied FVA, they did not automatically use current market prices for asset valuation. Many, if not most, assets recorded at FV were valued using models (or matrix pricing) even before the crisis. That is, banks relied on level 2 and level 3 fair values, often referred to as ‘mark-to-model’ accounting. However, the contagion effect of fair value accounting suggested by many theoretical models (e.g., Allen and Carletti 2008) builds on the assumption that fair values are current market prices. It is not obvious that this effect survives in the absence of this strong assumption.
  3. There were a number of safeguards built into the accounting rules, which acted as circuit breakers for downward spirals and contagion. For example, banks can move from using market prices to models when the liquidity in asset markets dries up, and many banks did so for their troubled assets in the crisis (Laux and Leuz 2010). Moreover, the accounting rules explicitly state that securities must not be valued using fire-sale prices. The rules also use the construct of “other-than-temporary impairments” to shield banks’ net income from the effects of “temporary” declines in the FV of their assets. In addition, many countries had prudential filters in place that shield banks’ regulatory capital from FV losses on available-for-sale (AFS) debt securities (Bischof et al. 2021). These safeguards weaken the links that are central to the potential problems with FVA.
  4. There is little evidence that banks systematically engaged in fire sales. In fact, the evidence suggests that commercial banks bought rather than sold mortgage-backed and other asset-backed securities during the crisis (e.g., Abbassi et al. 2016). One of the few examples of an alleged asset fire sale by a bank is Lone Star’s purchase of the Merrill Lynch CDO portfolio for 6.5 billion dollar or 22 cents on the dollar in July 2008. However, there is no indication that Merrill Lynch’s financial accounting triggered the sale or that the transaction (fire-sale) price became a relevant mark for other banks.
  5. In the aggregate, banks’ FV gains are not procyclical; expansionary periods are characterized by unrealized losses on AFS securities, not gains (Xie 2016). Moreover, there is no evidence that unrealized gains and losses on AFS securities under FVA are associated with procyclical leverage by U.S. banks (Laux and Rauter 2017). Complementary research on insurers also concludes that FVA did not magnify their problems, but instead induced (some) insurers to take actions earlier. In contrast, historical cost accounting is associated with gains trading (Ellul et al. 2015).

Other accounting problems were much more critical during the crisis

The heavy focus on FVA early on in the public debate (Bischof et al. 2020) has diverted attention from other accounting and reporting issues that also matter for financial stability. In fact, several early failures were banks that reported almost all of their assets at amortized cost, and hardly used FVA at all (e.g., IKB and Northern Rock). Our paper highlights the following issues:

  1. Banks disclosed their loss expectations very late and often only after the market had already learnt about them from other sources.
  2. Banks delayed the recognition of their loan losses. Bank managers’ incentives to avoid loss recognition seem to have played a larger role than the accounting rules.
  3. Regulators use prudential filters to shield banks’ regulatory capital from accounting losses with the intention of increasing stability in times of a crisis. However, the filters often dampened incentives for bank managers to take prompt corrective actions early during a crisis, which could undermine the stability effects.

Loss disclosures came very late and banks delayed the recognition of loan losses

The role of disclosures for financial stability is a priori not obvious. Regulators often stress the importance of disclosures for market discipline. However, investors could also overreact to negative news in a crisis, triggering bank runs. We therefore study the role of bank disclosures in the crisis. Our analysis documents that banks’ disclosures about relevant risk exposures came late, relative to the evolution of the crisis. Many banks apparently were reluctant to communicate their losses. Looking at credit default swap (CDS) spreads and share prices, we do not find strong adverse market responses to banks’ initial disclosures of subprime exposures. At the same time, these disclosures did little to resolve the large uncertainty in the market. Instead, relatively vague disclosures and repeated upward revisions in banks’ exposures and losses were prevalent. Thus, the problem was not that investors overreacted to bank information. Instead, it seems that poor and unreliable disclosures by the banks led to an erosion of trust, which likely was a bigger problem.

In addition to disclosures, banks’ loan loss recognition was frequently criticized, and the prevailing accounting rules were seen as a key reason for this “too-little-too-late” problem. Consistent with this policy debate, we find that banks’ impairments came late and were initially much too small. However, we do not find that the incurred loss model ‘constrained’ banks’ loss recognition. Instead, more likely explanations are banks’ reporting incentives, i.e., the reluctance by managers to recognize losses, and weak enforcement. These results raise the question of whether the new expected credit loss model under IFRS 9 and US-GAAP will have the intended effects, considering its forward-looking nature gives even more discretion to banks.

Prudential filters can harm financial stability because they countervail prompt corrective actions

Prudential filters shield regulatory capital from accounting losses. They can thus mitigate the threat of downward spirals. However, they also reduce banks’ incentives to take early corrective actions such as reducing dividends, raising equity, or selling securities that deteriorated in value. We find that the association between FV losses and banks’ corrective actions is weaker in countries where filters shield banks’ regulatory capital from FV losses, which is consistent with the concern that filters weaken incentives for corrective actions. Thus, the interaction between accounting and bank regulation deserves careful attention. Very often, ex post measures, intended to improve financial stability when a crisis is in full swing, have ex ante side effects with respect to banks’ actions (at which point the crisis could perhaps still be averted). Thus, regulators face thorny problems on the interaction of accounting and financial stability when regulating banks and more research on these interactions is needed. The COVID-19 crisis provides an opportunity to study several of these interactions, including the role of regulators.

Jannis Bischof is Professor of Accounting at the University of Mannheim. He is Principal Investigator of the TRR 266 Accounting for Transparency and the German Business Panel. He is also a Fellow of the Leibniz Institute for Financial Research SAFE and Goethe University’s Center for Advanced Studies on the Foundations of Law and Finance.

Christian Laux is Professor of Finance at WU (Vienna University of Economics and Business), member of the Vienna Graduate School of Finance (VGSF) and an ECGI Research Associate.

Christian Leuz is the Joseph Sondheimer Professor of International Economics, Finance and Accounting at the University of Chicago’s Booth School of Business and SAFE Fellow. He is a Research Associate at the National Bureau of Economic Research and a Fellow at the Center for Economic Policy Research, at the European Corporate Governance Institute, Goethe University’s Center for Financial Studies, and the CESifo Research Network.

Blog entries represent the authors‘ personal opinion and do not necessarily reflect the views of SAFE or its staff.