1 Introduction

The financial sector in 2007 had outgrown the post-depression protections which had been built to avoid another large financial crisis. As a result, governments had to intervene massively in their financial sectors, with limited tools and few guidelines. In The Power of Inaction, Cornelia Woll explores the intervention decisions in six countries to better understand how and why their interventions differed. It seems likely that we will encounter future crises for which our existing institutions are inadequate – a combination of innovation in the financial sector and gradual complacency among regulators have made financial crises endemic to modern capitalism (Roubini & Mihm, 2010). Almost by definition, the next crisis will have avoided our institutional safeguards, and so a better understanding of what we did in this crisis can help in intervening more effectively in such a future crisis.

Woll demonstrates that traditional political-economy categorizations (liberal market economy, coordinated market economy, small open economy) do not explain the different approaches taken in the cases studied, nor do arguments about the lobbying power of banks. Instead, Woll shows that the level of organization in the banking sector, and in particular, the extent to which it can be credibly claimed to be incapable of coordination, goes a long way to explain the difference in the level of private-sector participation and liability in the bailouts.

This comment is divided into three parts. First, a slight extension of Woll’s framework is offered by identifying cases of attempted inaction by government actors and the impact this had on two intervention episodes in the U.S., and one intervention in Germany. Second, the significance of the capacity for collective action among banks is critiqued, as it appears that the relative stability of a country’s largest bank(s) is sufficient to determine whether the main bailout package in the country was coordinated or not. Third, an unexplored implication of Woll’s findings is highlighted: employing a “Game of Chicken” framework ^[1] may explain the main shapes of the bailouts, but when one considers the costs these have resulted in, it appears that employing such an adversarial framework is ultimately counterproductive for the negotiating parties.

2 An extension: considering the other side of the table

The main thesis of the book is that banks are effectively engaged in a negotiating Game of Chicken with their government. They can avoid paying for the required interventions by convincing government decision-makers that they are incapable of coordination amongst themselves. This leads to an uncoordinated intervention in which the state shoulders nearly all the risk and potential costs. This is most clearly shown in the cases of the UK and Ireland, where banks had limited formal and informal sectoral coordination, and interventions were thus heavily state-led. ^[2] Coordinated interventions were possible in Denmark and France, where tight networks and a history of coordination prevented banks from feigning an inability to coordinate. In the U.S., banks had shown halting ability to coordinate, and so the final intervention was also semi-coordinated, involving broad-based participation on similar terms, but mostly due to government largesse on said terms of recapitalization (Woll, 2014, pp.98–101).

A closer look at bank coordination in the U.S., particularly in the weekend of the failed rescue of Lehman Brothers, shows us that the framework provided by Woll can be usefully extended by incorporating inaction by government actors as a counter to industry inaction. This government side of the negotiating table is part of the explanatory framework in The Power of Inaction, but has been mostly left out of the case studies therein. This is likely in part due to the explanatory power of the industry side, but a fuller framework helps in understanding some of the nuances in the crisis interventions.

Coordination between private banks was induced by intransigence among government actors during the weekend attempt at saving Lehman Brothers. The U.S. treasury secretary, Hank Paulson, had come under a lot of political pressure to be tough on banks following the massive bailout of Fannie Mae and Freddie Mac, and was also concerned about leaving a legacy: to his fellow regulators, Paulson “declared that he didn’t want to be known as ‘Mr. Bailout’.” (Geithner, 2014, pp.156–157). Paulson “didn’t like being the public face of Wall Street bailouts” (Bernanke, 2015, p.256), and as a result wanted to avoid bailing out the next bank. When Lehman Brothers needed saving, the regulators therefore called together the heads of the major banks to coordinate a solution. He told them, and leaked to the press, that there would be no government money involved in the rescue (Paulson, 2013, p.187; Bajaj, 2008; Craig, Mollenkamp, Solomon, & Fitzpatrick, 2008). ^[3] It is also possible that the looming collapse and likely bailout of AIG increased the pressure on government actors, further steeling them against expending political capital on saving Lehman Brothers. ^[4] Unfortunately, this determination was one of the main reasons Lehman Brothers declared bankruptcy on September 15^th, 2008. The accounts of both Paulson and Geithner point out that the assembled banks were willing to support a takeover of Lehman Brothers by sharing the risk between them for some of its worst assets (Paulson, 2013, p.206, 210; Geithner, 2014, p.161). To save the bank, however, they would still need a buyer, and none was forthcoming without government guarantees. The refusal of the U.S. to guarantee a US-based financial institution was one of the motivations behind the UK Chancellor rejecting a proposed merger between Lehman Brothers and Barclays (Darling, 2011, pp.122–123). It was known that the blow from Lehman would be hard, but it was also believed to be “manageable”, and so the government opted for inaction in spite of limited but coordinated action by the assembled banks. In Woll’s account, this is cited as an example of the unwillingness among US banks to “link their fortunes through a joint rescue plan” (Woll, 2014, p.101), but as the banks were willing to support a deal, it is more accurately portayed as an example of the government being unwilling to commit to a joint rescue plan. Furthermore, in the face of government inaction, and in response to the now unavoidable bankruptcy of Lehman Brother, assembled bank did in fact agree to link their fortunes: at the end of the weekend, 11 major international banks agreed to establish a shared liquidity pool worth $70bn, which was to help them through the turmoil (Scholtes & Guha, 2008; Clark, 2008).

The main decision-makers in the U.S. bailouts, Paulson, Geithner, and Bernanke, quickly reconsidered their anti-bailout stance in the wake of the Lehman Bankruptcy and bailed out American International Group (AIG) the day after Lehman Brothers declared bankruptcy. ^[5] However, they now sought additional funds and authority from Congress, and congressional representatives were predominantly against bailouts of banks: “Washington had become a cauldron of Old Testament populism and moral hazard fundamentalism.” (Geithner, 2014, p.156), in part because, as representative Barney Frank allegedly put it, “No one will ever get reelected for avoiding a crisis.” (Paulson, 2013, p.261). The animosity led the House of Representatives to roundly reject the Emergency Economic Stabilization Act when it was put to a vote, even as major banks and entire markets had been failing on a nearly daily basis. The revolt was, however, short-lived: the bill passed a few days later, following a further dramatic market collapse and the addition of tax “sweeteners” for opposed representatives: “The change in course by the House was prompted by fears of a global economic meltdown, and by old-fashioned political inducements added by the Senate: a portfolio of $150 billion in popular tax provisions.” (Herszenhorn, 2008; Hitt & Lueck, 2008; Hulse & Pear, 2008).

The relatively slow and stumbled passage of the bill served to strengthen the negotiating position of government actors, vis-à-vis banks, in the subsequent intervention. When the U.S. regulators summoned the banks and offered blanket recapitalizations, the broad participation was mainly the result of the favourable terms of the recapitalizations (Woll, 2014, p.101), ^[6] but to this can be added that government inaction strengthened its negotiating position and made banks more willing to accept the terms. The prolonged negotiations in Congress had allowed the positions of banks to worsen, with most of the major banks either scrambling to raise capital and liquidity (Goldman Sachs and Morgan Stanley) or taking on additional risk to save other failing banks (JP Morgan, Bank of America, and Wells Fargo). ^[7] This made defection on the part of banks more costly when they finally met in mid-October. Hank Paulson summarized their situation in an interview: “I don’t think there was a banker in the room that was going to look us in the eye and say they had too much capital.” (Landler & Dash, 2008). As a result, the meeting between the banks and the regulators did not last long: Wells Fargo’s CEO was opposed at first, and there was some bickering on details, but after a short hour, the reality of the banks’ negotiating position was summarized by Bank of America CEO, Ken Lewis: “We’re all going to do this. Let’s just get it done.” (Wessel, 2010, p.239; Landler & Dash, 2008). The uncertain capital positions of the banks, and the cheap price of the offered capital, quickly undermined any attempts at negotiating a better deal from the government.

Government defection spurred bank coordination in the Lehman Brother’ weekend (albeit not enough), and possibly contributed to coordination in the implementation of the main recapitalizations in the U.S. Woll recounts a similar pattern in the German rescue of Hypo Real Estate (HRE), but did not emphasize the relative success of government inaction in this example (Woll, 2014, pp.129–130). The German Ministry of Finance (MoF) briefly tried to defect from the rescue of HRE and obtained significant commitments from the private sector by doing so. Although, on the surface, there had been a series of coordinated bailouts in Germany since the summer of 2007, HRE was the first one to fail in which the government did not have any ownership stake, and so the government was not automatically obligated to save it. ^[8] When the HRE was collapsing at the end of September 2008, the Minister of Finance, Peer Steinbrück, called together the heads of the private bank so that they could coordinate a rescue, and then deliberately stayed away from the talks for two and a half days, during which there was virtually no communication between the negotiating banks and the MoF (Balzli et al., 2008). This tactic was explicitly acknowledged by government representative Jens Wiedmann: “our assessment at the time was that the sooner a government representative participated in the negotiations, the sooner claims would be made on the federal government. We wanted to maximize the contribution from the financial sector.” ^[9] (Krüger et al., 2009, p.135, my translation). After these days of unsuccessful negotiations by private banks, the MoF reluctantly sent a representative and a joint public-private bailout was agreed on. Notably, however, the resulting rescue involved only guarantees from the state, with all the required funding provided by the private banks and the European Central Bank. The German experiment with defection was, like the U.S. ones, short-lived: the outcome of the HRE rescue convinced government decision-makers that there were not enough funds in German banks for the system to save itself. In contrast to previous rescues in Germany, the main German bailout package was state-led and voluntary for banks, leading to an initially high stigma of participation and eventually large losses for the state (Bundesanstalt für Finanzmarktstabilisierung, 2014).

The above examples show that, on occasion, governments did act as an active player in the negotiations and that government decisions to cooperate or defect have noticeable effects on the outcomes. A fuller analysis would be needed, however, to fully understand when and why governments attempt to defect.

3 A Critique: The Problem with Over-determination and Small-n Inferences

When you have a limited number of cases, it can happen that several factors can explain the same observational outcomes, leaving it unclear which were the most important. In the final chapter of the book, Woll offers an elaboration of the causes of financial sector inaction in the six countries, but this elaboration amounts to much simplified sufficient explanation: coordinated interventions with industry involvement were possible in countries where the largest and most significant bank or banks were in trouble and was not possible in countries where these banks did not need help (Woll, 2014, p.117). This explanation presents a problem for the significance of a capacity for industry coordination insofar that it can, on its own, explain the shape of the main bailout in all the cases, even without knowing anything about the history or pre-existing capacity for coordination in the countries’ financial systems. Coordinated bailouts were possible in the U.S., France, and Denmark because nearly all the major banks in the U.S. and France, and the domineering Danske Bank in Denmark, were in trouble. In the UK, however, Barclays and HSBC were not sufficiently in trouble for them to have an interest in risk-sharing (and RBS had only become the largest bank the previous year by swallowing up ABN Amro). In Germany, Deutsche Bank was sufficiently secure in its position that it could get by without direct aid from the German government, in part due to receiving nearly $12bn from the U.S. bailout of AIG (Steinbrück, 2010, p.206). Its non-cooperation made the main German bailout package a purely government-run, and stigmatizing, affair. Finally, in Ireland, the two largest banks requested a bailout from the government, as they were heavily reliant on wholesale funding, but real losses, as opposed to liquidity constraints, would have been concentrated at the smaller Anglo-Irish Bank and Irish Nationwide Building Society, as these were more heavily exposed to the real estate sector (Woll, 2014, p.144). This makes it understandable that the largest Irish banks did not pursue a coordinated rescue or any sort of cross-subsidization.

It remains plausible that the pre-existing organization of the banking sector, its history, and other factors affect the capacity for collective (in)action among banks, are relevant factors which could affect negotiated bailout outcomes. But insofar that having the largest banks in trouble appears to be sufficient to explain the bailout process outcome, we can then wonder how relevant other factors were. The forms of the main bailouts were overdetermined, and so with limited cases, unfortunately, we cannot draw firm conclusions – at least not from pattern-matching.

4 An Implication: Is a Game of Chicken the Right Framework?

A Game of Chicken is a nearby framework for approaching the bailout negotiations between the state and private banks: the non-cooperating party stands to gain more (or at least lose less) if the other party cooperates, but both face a large cost if neither cooperate and intervene in the crisis. Its explanatory power as an analytical framework is a result of bank and government actors actually approaching their negotiations from the standpoint of this framework. However, the traditional Game of Chicken makes fairly stringent assumptions about the payoff matrix, which are not reflective of reality of intervening in a financial crisis. In particular, the symmetric nature of the payoff matrix, as exemplified in Table 1, assumes that both parties have similar capabilities and face similar choices. In addition, the game of chicken generally assumes that there are no cooperative gains – meaning the sum of payoffs in Cell A in Table 1 is equivalent to the sums of payoffs in Cells B and C (Snyder, 1971). This section uses evidence from The Power of Inaction and other sources to show the error of these assumptions and elaborate on their implication for the negotiating position of governments.

This payoff matrix does not conform to what we have learnt about bank bailouts, and must be modified in at least two ways. First, in a systemic banking crisis, it is often not realistic to assume that the banks could save themselves or each other if left to their own devices. This may be possible if just one or two banks were in trouble, but in a systemic crisis, there are not enough resources in the banking sector to resolve the crisis. Eventually, private banks run out of the capital buffers needed to absorb other banks, and potential private-sector investors face a collective action problem whereby investing in falling bank stocks during a crisis is only a good idea when everyone else does it. Because banks often mirror each others’ exposures, losses in a banking crisis are likely to affect many banks simultaneously, leaving few to save the sector (Acharya & Yorulmazer, 2007; Luengnaruemitchai & Wilcox, 2004; Guttentag & Herring, 1985, p.136). Crises themselves are often the result of many banks competing to pursue profits from a poorly understood financial innovation (Hautcoeur & Riva, 2013; Boyer, 2013). The limits of private-sector solutions were reached with the HRE bailout in Germany (Woll, 2014, p.130) and to some extent during the Lehman Brother’s weekend in the U.S. when the seemingly only U.S. banks with sufficient headroom to absorb Lehman Brothers decided to target Merrill Lynch instead. The limited capacity for a state-less crisis response in the US is further highlighted by the merged Bank of America/Merrill Lynch company’s need for an additional $20bn in TARP funds, over and above the $25bn they initially received (Financial Crisis Inquiry Commission, 2011, pp.382–386). Although a negotiation about a bailout can be expected to create an adversarial framework about the distribution of costs, the effect of this change is that the state faces a loss if it defects, even if banks cooperate. Cell A becomes favorable to cell B, meaning that the state receives a better return (or has lower costs) if it cooperates, irrespective of whether banks cooperate or defect.

The second adjustment is that, contrary to a typical game of chicken, there may be collective gains to be made through mutual cooperation. Because financial crises thrive on uncertainty and weak market participants, a coordinated rescue between the banks and governments is likely to be more efficient, and therefore more likely to yield a positive economic return for participating parties. What this means is that the sum of values in cell A is greater than the sum of values in cell C (as well as the worsened cell B).

The evidence from Woll’s book and elsewhere indicates that coordinated interventions yield a positive fiscal return for the state. The uncoordinated German rescue package is still projected to result in losses of over €20bn (Bundesanstalt für Finanzmarktstabilisierung, 2014), and the uncoordinated UK interventions are, by an optimistic estimate, headed for a surplus (Rothschild, 2015), but this is without accounting for the interest the state has had to pay on the debt taken on to recapitalize banks. The uncoordinated intervention in Ireland was so expensive for the government that it later required its own bailout from its EU peers.

By contrast, the coordinated interventions in France and Denmark produced a surplus for the states (Woll, 2014, p.8, 12) and the coordinated recapitalizations in the U.S. provided a positive return to the state of over $20bn (before interest), making it one of the most profitable parts of the overall U.S. intervention (Kiel & Nguyen, 2015; Department of the Treasury, 2010, 2014). ^[10],[11]

It is not apparent from the cases that banks as a group profit from a coordinated intervention, although it is possible – the costs for banks were not exactly cheap in the UK and Ireland, where there was no coordination (Woll, 2014, p.173). Assuming they do not profit much, and that they carry some or the costs of the coordinated intervention, we should expect the banks’ payoff in A to be less than their payoff in C, meaning that the positive-sum gains from moving to a coordinated intervention flow disproportionately to the state.

This payoff matrix is exemplified in Table 2. One can quibble over the exact magnitudes of the numbers chosen, ^[12] but what matters are the assumptions highlighted above: a mutually poor outcome in cell B, and a positive-sum outcome in cell A with the benefit for the state (relative to cell C) greater than that of the banks.

In this arrangement, banks are still incentivized to defect, but the government can change the their behavior by providing side-payments which would make up for the difference in their gains from Cell C to A. Because the government receives a disproportionate share of these cooperative gains, it can use some of them to essentially bribe banks to cooperate and still end up in a better financial position at the end, much like the U.S. government actors did when it offered major banks recapitalizations at a very low interest rate. In theory, institutionalizing such side-payments would make future cooperative interventions much easier to achieve and as a result reduce uncertainty and market panic. Negotiating parties naturally employ an adversarial framework when determining the distribution of costs for an intervention, but given the above evidence, acting as if the outcomes resemble a Game of Chicken is counterproductive and may lead to lower-sum outcomes.

Woll advocates for the government to take an adversarial negotiating stance and prefers the imposition of a unilateral government plan even if the government could reduce costs by sponsoring a coordinated intervention (Woll, 2014, p.63). However, Woll also acknowledges that this is a somewhat normative prescription and that it is based on the assumption that government-sponsored coordination would increase moral hazard (Woll, 2014, p.63). More evidence would be needed to substantiate or rebuke this assumption, but on a theoretical level it is worth noting that government bailouts primarily induce moral hazard among the creditors of the rescued banks (Stern & Feldman, 2004). Without better evidence, future government negotiators should at least consider the savings which can be made by fostering coordinated interventions.

5 Conclusion

The Power of Inaction provides both a comprehensive and informative empirical reconstruction of the bank bailouts in six country case studies and is a valuable first attempt at explaining the causes and consequences of particular intervention decisions in the Global Financial Crisis in particular, and in financial crises more generally. This comment has sought to both extend and critique this explanation.

In terms of extensions, there is some evidence from the U.S. and Germany that the threat of defection can be wielded by the government as well as the banks in bailout negotiations, leading to more industry coordination and possibly better terms for the state. That said, such a government stance is not necessarily wise, as further evidence also indicates that a joint public-private intervention produces a better outcome for the state and the taxpayer, than does a state-led or industry-led intervention, the latter of which is generally unviable. Rather than wield the threat of defection, government actors will likely find a better bailout outcome if they persuade uncooperative banks to participate in a coordinated rescue.

In term of critiques, one concern was raised: given that coordinated interventions were staged in countries where the largest bank(s) were in trouble, and uncoordinated interventions were staged in countries where the largest bank(s) were relatively stable, there is little room to make inferences concerning the more general capacity for coordination within a country’s banking industry. Given that the above factor is sufficient in determining outcomes, it is possible that the history and internal structure of a country’s banking sector matters little, or not at all.