Financial Stability and Credit Policy

Douglas Diamond and Philip Dybvig published "Bank Runs, Deposit Insurance, and Liquidity" in the Journal of Political Economy in 1983 ^[1]. The model is two periods, two types of agents, and one bank. Patient agents value consumption in period two. Impatient agents value consumption in period one. The bank takes deposits, invests in a long-duration asset, and promises every depositor the option to withdraw at par on demand. The model has two equilibria. In one, only impatient agents withdraw early; the bank pays them out and the patient agents wait for the higher long-period return. In the other, everyone withdraws because everyone fears everyone else will withdraw, and the bank is forced to liquidate the long asset at fire-sale prices.

The point of the model is that the second equilibrium is not irrational. It is a Nash equilibrium given the bank's promise structure. Diamond and Dybvig won the 2022 Nobel for showing that deposit insurance picks the good equilibrium for free, at the cost of moral hazard. Without insurance, banking is unstable. With it, banks have to be regulated because the insurance guarantee changes the price of risk.

The 2008 collapse and the 2023 SVB case are both Diamond-Dybvig stories with different runners. In 2008 the runners were institutional money-market funds and repo lenders, not retail depositors. The Reserve Primary Fund broke the buck on September 16, 2008. The Fed's response was to extend the lender-of-last-resort backstop to wholesale funding markets through the Primary Dealer Credit Facility and the Money Market Investor Funding Facility ^[2]. In 2023 the runners were uninsured tech-sector depositors at Silicon Valley Bank coordinating through social channels. The FDIC's response was to invoke the systemic-risk exception and guarantee uninsured deposits. Both cases were the model in operation. The institution that ran was different. The structural problem was the same.

Basel I (1988) set a single requirement: tier 1 capital had to be at least 4 percent of risk-weighted assets, with total capital at 8 percent. The risk weights were crude. Sovereign debt of OECD countries got zero weight. Mortgages got 50 percent. The framework was negotiated by the Basel Committee on Banking Supervision and signed by the G10 supervisors. Implementation was domestic, which is why the rules came into effect in different years across countries.

Basel II (2004) replaced the standardised weights with internal-models-based estimates. Banks could use their own value-at-risk calculations to compute capital charges. The intent was risk-sensitive regulation. The realised effect was capital arbitrage: the banks with the largest balance sheets generated the lowest model-based weights, which let them grow leverage in 2005-2007 with regulatory blessing. The crisis made clear the framework was procyclical and gameable.

Basel III (2010, phased in through 2028) added three constraints to the risk-weighted ratio: a leverage ratio that does not depend on internal models, a liquidity coverage ratio that requires high-quality liquid assets sized to cover 30 days of stressed outflows, and a net stable funding ratio that prevents banks from funding long-duration assets with overnight wholesale liabilities ^[3]. The countercyclical capital buffer was added as a discretionary tool. The framework also raised the risk-weighted minimum to 7 percent for tier 1 plus a 2.5 percent capital conservation buffer, with global systemically important banks facing surcharges on top.

The 2023 SVB case showed the limits of Basel III's coverage. Mid-sized U.S. banks, defined under the 2018 EGRRCPA carve-out as institutions below $250 billion in assets, were exempted from the LCR and from the strictest stress tests. SVB was one of those banks. Its balance sheet was concentrated in long-duration Treasuries that lost significant market value as rates rose. Held-to-maturity accounting masked the loss. The bank was Basel-III compliant on paper. The framework as implemented in the United States did not catch what the framework as designed by the Basel Committee was supposed to catch ^[4].

The conventional periodization splits the 2007-2009 crisis into three phases. Phase one ran from August 2007, when BNP Paribas froze redemptions on three structured-credit funds, to September 2008. The Fed's response was a sequence of new lending facilities aimed at specific market dysfunctions: the TAF for term funding, the TSLF for primary dealers, the PDCF for broker-dealers ^[5]. The federal funds rate fell substantially from its mid-2007 level by April 2008.

Phase two began on September 7, 2008, when Treasury placed Fannie Mae and Freddie Mac into conservatorship. Lehman filed for Chapter 11 on September 15. AIG was rescued on September 16 with an initial credit line that was expanded substantially over the following months. The Reserve Primary Fund broke the buck on September 16. TARP was passed on October 3. The fed funds rate reached zero by December 16. The first round of large-scale asset purchases was announced on November 25, initially targeting agency MBS and agency debt.

Phase three was the slow recovery. The output gap closed years after the recession's trough, and the Fed did not raise rates until December 2015. The fiscal response, ARRA, contributed stimulus that was substantially smaller than the CBO's estimate of the peak output gap. The mismatch is why the recovery took as long as it did and why hysteresis effects such as lower labor-force participation and slower productivity growth became part of the empirical record ^[6].

The lasting institutional changes were Dodd-Frank in the United States (July 2010) and the Single Supervisory Mechanism in the euro area (November 2014). The Federal Reserve gained section 13(3) emergency-lending authority that was substantially curtailed by Dodd-Frank in exchange for the Treasury's pre-positioned guarantee. The Volcker Rule banned proprietary trading by deposit-taking institutions. The Financial Stability Oversight Council was created to designate non-bank systemically important institutions, though only a handful were ever designated, and all designations were lifted by 2018.