Thank you to the following individuals for their thoughtful comments on the text: Andrew Metrick, Chase Ross, Ted Truman, Lee Sachs, Meg McConnell, Matt Kabaker, and Jake Siewert.

PBS Frontline (1997), Interview with Rudi Dornbusch.

See, for example, Gorton (2016), “The History and Economics of Safe Assets,” NBER Working Paper.

Levine (2016), “Regulators Want to Slow Runs on Derivatives,” Bloomberg.

Summers (2010), “Financial Stability: Retrospect and Prospect,” remarks at the Stanford Institute for Economic Policy Research.

FDIC (2016), Bank Data & Statistics, Historical Statistics on Banking, Table CB14, “Liabilities and Equity Capital.”

Bao, David, and Han (2015), “The Runnables,” FEDS Notes.

For a time series of net repo funding to broker/dealers and banks based on flow of funds data, see Gorton and Metrick (2015), “Who Ran on Repo?” NBER Working Paper. See also Buehler, Noteboom, and Williams (2013), “Between Deluge and Drought: The Future of US Bank Liquidity and Funding,” McKinsey Working Papers on Risk.

Baklanova, Copeland, and McCaughrin (2015), “Reference Guide to U.S. Repo and Securities Lending Markets,” Federal Reserve Bank of New York Staff Reports, Figure 9.

“Capital requirements for banks are much higher, as are risk weights and the quality of bank capital. In all, new capital requirements are at least seven times the pre-crisis standards for most banks. For globally systemic banks, they are more than ten times.” Carney (2014), “The Future of Financial Reform.”

FDIC (2016), Historical Statistics on Banking, Table CB14.

IMF (2013), Global Financial Stability Report.

For additional details, see Adrian and Ashcraft (2012), “Shadow Banking Regulation,” Federal Reserve Bank of New York Staff Reports.

For example, Gorton, Lewellen, and Metrick (2012) show the components of safe assets changed over a 30-year time window, with bank deposits constituting 70 percent of the safe-asset share in the 1970s, but falling to 27 percent before the financial crisis as money market mutual funds, broker-dealer commercial paper, securitized debt from GSEs and other asset-backed securities (ABS) grew.

Duca (2016) examines the linkage between higher capital requirements and the shadow bank share of short-term business credit in the U.S. over many decades. At a shorter horizon, Xie (2012) documents the increase of ABS/mortgage-backed securities (MBS) and asset-backed commercial paper (ABCP) issuance on a daily basis when expected convenience yield is high. Sunderam (2015) finds a similar phenomenon on a weekly basis, which suggests investors regard shadow bank debt as money-like.

Sections 23A and 23B of the Federal Reserve Act “limit the risks to a bank from transactions between the bank and its affiliates and limit the ability of a bank to transfer to its affiliates the subsidy arising from the bank’s access to the Federal safety net.” Section 23A identifies eligible transactions between a bank and any single affiliate of the bank, and Section 23B requires that certain transactions between a bank and its affiliate occur on market terms. The Board of Governors of the Federal Reserve may waive these requirements, by a vote of the governors, if such an action would be in the public’s best interest. See Board of Governors of the Federal Reserve System (2003), “Adoption of Regulation W Implementing Sections 23A and 23B of the Federal Reserve Act.”

Lenza, Pill, and Reichlin (2010) compare the Federal Reserve’s and European Central Bank’s (ECB’s) initial interventions during the global financial crisis in the context of their institutional design differences. With its foundation in the pre–Monetary Union period, the ECB could regularly transact with a wider set of counterparties and securities, which notably included ABS. Pre-crisis, almost 2,000 credit institutions could participate in weekly ECB operations, compared to the U.S. system with a few dozen primary dealers which participated in regular daily operations. As a result, the ECB’s main refinancing operations before the crisis averaged €300 billion, whereas the Fed’s routine refinancing operations averaged about $30 billion.

See Labonte (2016) for a discussion of the notable changes to the Federal Reserve’s Section 13(3) authorities. Some of the most important changes: Section 13(3) assistance must now be broad based, meaning at least five eligible participants meet the eligibility requirements; provision of liquidity can only be to an “identifiable market or sector of the financial system”; assistance requires the approval of the Secretary of the Treasury; and information on borrowers must be provided to relevant congressional committees within seven days.