The Decline of Too Big to Fail

8 Jan 2020
Read Time 3 mins
Categories :

Whether large banks are “Too Big to Fail” (TBTF) is an important public policy issue that is being seriously considered by policymakers.  Indeed, the Financial Stability Board (FSB) is currently engaged in an ongoing project to evaluate the effectiveness of TBTF reforms that have been enacted over the past decade.  Recently, researchers at Stanford University and Australian National University released a research paper, “The Decline of Too Big to Fail,” that offers fresh, data-based insight into this important question.  Specifically, the researchers study whether investor expectations of a bailout in the event of a large bank’s default have changed over the past decade and find that bailout expectations have declined significantly. The extent to which bailout expectations or market-implied probabilities of government bailouts have changed is important because it speaks directly to whether TBTF-related reforms have been effective.

In this post, we summarize the researchers’ main findings and offer thoughts on how the FSB can incorporate this research into its evaluation.

In their paper, the researchers use the idea that the market-implied probability of a firm’s bailout can be measured using the simple relationship S = p x L x (1 – b), where represents the corporate credit spread or the cost of borrowed funds,  p represents the probability of default, L represents the loss in the event of default when no bailout occurs, and  b represents the probability of a bailout. This relationship implies that an increase in a bank’s cost of debt (S) that is not accompanied by an increase in the bank’s default risk (p x L) signals a decline in investors’ bailout expectations (b).  As investors’ bailout expectations recede, they demand a higher interest rate on funds they lend to the bank to compensate for the greater risk of loss in the unlikely event of default.  This simple insight is the key to understanding the researchers’ empirical findings. 

Using the above relationship, the researchers estimate a structural dynamic model of debt and equity prices to compare market-implied bailout probabilities for large U.S. banks before and after the default of Lehman Brothers in September of 2008.  The researchers calibrate their model using a dataset comprised of debt costs for a range of large U.S. banks and other firms from 2002 to 2017. 

The researchers’ main findings are summarized by Figure 3 of the paper, which is reproduced below.  The blue line in Figure 3 shows the credit spread or “CDS” spread of a representative large bank with a given level of riskiness.  The dashed red line beginning at the date of the Lehman failure in 2008 shows the level of the credit spread that would have prevailed had there been no change in investors’ expectations of a bailout.  The difference between the two lines is a measure of the extent to which large bank debt costs have changed as a result of changing bailout expectations.  Looking at the figure shows that since 2008, the credit spread paid by large banks to borrow money from investors has increased by roughly one percentage point as a result of a decline in investor bailout expectations.  Put differently, if investors maintained the same expectations for a government bailout that they held before 2008, large banks would have paid one percentage point less to borrow money from debt investors.  Given the average level of interest rates of one to two percent over the past decade, a one percentage point change in debt funding costs is significant.

Source: Berndt, Antje and Duffie, James Darrell and Zhu, Yichao, The Decline of Too Big to Fail (December 1, 2019). Available at SSRN: or

So why have debt costs risen for large banks?  The researchers conclude that policies aimed at addressing TBTF for globally systemically important banks (G-SIBs) have been successful.  Specifically, they argue that

the data are consistent with significant effectiveness for the official sector’s post-Lehman G-SIB failure resolution intentions, laws, and rules.  G-SIB creditors now appear to expect to suffer much larger losses in the event that a G-SIB approaches insolvency.  In this sense, we estimate a major decline in ‘too big to fail’.”

Of course, no single study is conclusive, but as we have previously profiled, a large array of academic evidence strongly suggests that the problem of TBTF has been effectively addressed by policies enacted over the past decade. Policymakers should consider this evidence, including the new and convincing analysis in this new research paper, as they evaluate the effectiveness of TBTF policies.           

Forum Updates