[This is part one of a three part series on the financial crisis and banking.]

An insolvent bank is not the same thing as an illiquid bank. Insolvency has to do with an imbalance between all assets and liabilities; illiquidity refers to the inability to meet short-term obligations, because the firm cannot quickly sell some of its relatively illiquid assets at par (to generate the sufficient liquidity). Systemic banking crises, such as the classic bank run, usually have more to do with illiquidity than insolvency. Noah Smith writes a bit about illiquidity versus insolvency in the context of the 2007–09 financial crisis. But, Noah distinguishes the two by drawing a distinction between events that include a deterioration of the asset side of banks’ balance sheets and those that don’t. Can’t negative shocks to bank assets, however, create liquidity crises?

Noah distinguishes between two explanations of the recent financial crisis. The first is the one proposed by economists such as Gary Gorton and Andrew Metrick (ungated),¹ who argue that the crisis was a result of a run on investment banks through wholesale credit markets (e.g. repo). In a nutshell, wholesale funding allow banks to borrow large sums of money from agents with high cash savings, offering an asset — prior to 2007–08, this asset was typically the mortgage backed security or a derivative of it — as collateral. Wholesale credit is a very short-term loan that provides banks with the liquidity to carry out day-to-day operations. They can be described as large, short-term deposits that earn interest. Between 2002–07, wholesale credit markets were active and growing, but then collapsed during the crisis.² Gorton and Metrick argue that the housing market collapse caused concern about the health of banks’ assets, even those not directly related to housing, inducing a run on the banking system. Lenders were worried that banks would not be able to repurchase the posted collateral at full price, this collateral took a haircut, banks were unable to borrow on the same terms to repay their obligations, and the system suffered a liquidity crisis.

The second interpretation of the crisis is the one given by economists such as Paul Krugman and Anna Schwartz. This position is a little bit more ambiguous in the details. Schwartz argued that credit markets dried up, because investors were unsure if banks would be able to meet their obligations. Both Krugman and Schwartz allude to the deterioration of bank assets. But, there is a difference between maturity mismatch and the event where all of a bank’s assets are worth less than its liabilities, and a negative shock to part of a bank’s assets does not necessarily imply the latter. For Krugman and Schwartz to be right, there must have been a systemic imbalance between total assets and total liabilities.

One issue worth highlighting is that both sides may be getting at the same general point — the bank run —, and that the differentiation between insolvency and illiquidity is, to some extent, semantic. We can take an alternative approach and differentiate between “currency runs” and “redemption runs” (Selgin [1988]³, pp. 133–139). The former are cases in which depositors want to convert deposits into banknotes (or cash), and the latter describes events where customers want to redeem a bank’s liabilities for its assets (although, usually in liquid form). If a bank faces legal restrictions in issuing money, currency runs can lead to liquidity crises; redemption runs can lead to liquidity or insolvency crises. In any case, Gorton and Metrick, Krugman, and Schwartz all explain the 2007–09 crisis as a redemption run.

Coming back to the debate on illiquidity versus insolvency, the 2007–09 was most likely a liquidity crisis, but one that could have very well turned into a solvency crisis. While there were probably a number of banks that did suffer solvency issues, the question is whether insolvency was systemic.

It needs to be clarified that insolvency is not necessarily the same thing as bankruptcy, which is a legal definition of insolvency. Legally, bank solvency is related to its capital adequacy ratio, or the amount of equity, retained earnings, and certain debt instruments a bank holds in proportion to its total assets, which are weighed by measured risk. When the mortgage backed security market collapsed, the risk associated with these assets increased, forcing banks to scramble to meet the capital adequacy requirement. This caused banks to hoard liquidity, making it more difficult to meet short-term obligations (Friedman and Kraus [2011],⁴ pp. 91–96). Related, mark-to-market laws exacerbated the crisis, because the original downward revision of asset values was exaggerated; the actual losses over these bonds’ lifetime were smaller (and the revenue stream associated with these assets had not necessarily shrunk). But capitalization laws required banks to take the exaggerated loss when judging the solvency of their balance sheets (Ibid., pp. 97–100).

During the crisis, banks did not necessarily suffer an imbalance between total assets and total liabilities. The problem was one of maturity mismatching. Prior to the crisis, mortgage backed securities were considered low-risk, low-return assets, and they were relatively liquid. Banks could post these assets as collateral and borrow from wholesale depositors, using this liquidity to deal with day-to-day transactions. When the housing market turned, the risk-weighted value of these assets became uncertain, and they lost their liquidity. Banks were not able to turn them into cash, without taking an immediate loss that was much greater than the actual loss in the bonds’ values. Further, whatever liquidity banks could get their hands on was, in a way, tied down by the scramble to meet minimum capital adequacy requirements, even though it can be argued that meeting these minimum standards should not have been a priority. Banks were unable to meet short-term obligations with the liquid assets (e.g. cash) their depositors were demanding.

Why did the Fed buy $1.25 trillion worth of agency securities in 2009? The Fed was aware of these assets’ illiquidity, therefore they essentially built a new wholesale credit market, becoming their sole buyer. This interpretation is supported by Ben Bernanke’s 2009 lecture at the London School of Economics. Bernanke draws a distinction between quantitative easing and credit easing, where the former focuses on the banking system’s reserves, influencing the composition of banks’ assets indirectly. Credit easing, by contrast, is a program to specifically target bank assets and change their composition. The likely rationale behind the Fed’s program was that credit channels had deteriorated and the intermediation of liquidity had been consequently compromised, making it more difficult for all parties to access the liquidity necessary to meet short-term obligations (Bernanke [1983], pp. 263–268). We could ask whether the Fed paid too much for these assets, implying that part of the “credit easing” program was an implicit subsidy, but, with the above point on lifelong value of the bond in mind, this subsidy was probably not the difference between solvency and insolvency.

Most bank runs, whether of wholesale or commercial deposits, are usually precipitated by negative shocks to a bank’s balance sheet. Bank runs are not sunspot phenomena, they are usually induced by some prior event.⁵ Consider bank runs during the mid-19th century United States. Redemption runs on banks were typically caused by negative shocks to collateral. Free banking laws required banks to back their privately issued banknotes with state bonds, which were considered risk-less by regulators. As it turned out, these bonds were actually not risk-less, and when states defaulted on their debts it put into question the ability of banks to meet currency redemption demands, inducing bank runs (Gorton [2012], pp. 16–17). Similarly, there is reason to believe that the bank runs of the Great Depression were caused, at least in part, by a deterioration in the value of some of the banks’ assets, specifically loans (White [1984], p. 126–128).

We cannot use negative shocks to bank assets as a way of distinguishing between solvency and liquidity crises. The two are probably related in cause, and may represent different equilibria — depending on how depositors react to new information about banks’ asset values and the actual financial position of the banking system (whether a bank’s total assets are valued at least the value of its total liabilities). But, negative shocks to banks’ assets can occur without causing an insolvency crisis. The 2007–09 financial crisis was probably a systemic liquidity crisis, although there may have been a significant number of financial institutions which were also insolvent. AIG Financial Products (AIG FP) offers us an example that helps distinguish. AIG FP had accepted a large pool of liabilities in the shape of credit default swaps. When the crisis occured, this triggered a ballooning of the value of AIG FP’s liabilities, triggering bankruptcy. This liability-side cause of its bankruptcy stands in contrast to the asset-side cause of the run on wholesale deposits.

______________________________________

Notes:

1. If you are interested in bank theory and/or the financial crisis, I recommend reading Gorton and Metrick (2009). I don’t think there is anybody who can explain the bank run theory as clearly, and as detailed, as them. The final version of their paper, published by the Journal of Financial Economics, can be found here — I’m not sure how different this version is from the NBER working paper.
2. FRED has data on the total value of repurchase agreements and asset backed commercial paper.
3. A free version of George Selgin’s The Theory of Free Banking can be found here.
4. See my review of Friedman and Kraus’ Engineering the Financial Crisis.
5. This is an admittedly controversial issue. See, for example, Ennis (2003) for an argument in favor of the opposite conclusion. My challenge is that the existence of multiple equilibria — say, one where there is a bank run and another where there is no bank run — is not inconsistent with the idea that there is a causal relationship between some prior event and liquidity and solvency crises. Banking crises are complex events; there are typically many (mostly unpredictable, only observable) factors that contribute to its occurrence. Which factors are relevant (at some point in time) is an empirical question. But, depending on which factors are relevant, the “system” can end up at different equilibria. Multiple equilibria do not imply that bank runs are “random,” except maybe in a trivial sense of the word.