Recent weeks have witnessed several notable events on the systemic bank-restructuring front. In the US, following the conclusion of the stress tests, ten banks that were recapitalised under the TARP program have been granted the right to repay TARP funds, and some have already exited the program. At the same time, asset sales under the Public-Private Investment Program for Legacy Assets have been delayed and are starting at a much smaller scale than initially planned. In Europe, observers have called on governments to conduct stress tests for their banks and launch a resolute and coordinated clean up of the banking system. These events are evidence of continued shifts in approaches to systemic bank restructuring on both sides of the Atlantic. They reflect new data and developments, but they also reveal disagreement among observers, economists, and policy makers about the best way forward.
As Oliver Hart and Luigi Zingales outlined in the Wall Street Journal, in the midst of the financial crisis, rescue packages involving public funds ought to be designed using economic first principles, given the money at stake and the ability of lobbies to spin their wishes as common wisdom. The tools employed to rescue systemically important banks are often the same in cases of systemic banking crises. As mentioned by Michael Pomerleano on VoxEU, the standard prescription includes “a rigorous assessment of major banks’ balance sheets, removal of nonperforming loans from banks’ balance sheets, and banks’ recapitalisation”. Yet, these tools’ costs for the taxpayer have rarely been studied. What makes recapitalisation better or worse than asset sales? Why is insurance used in some circumstances but not in others? What tools are the best from the objective of restoring bank solvency at the lowest public cost?
In a recent paper, we review these issues (Landier and Ueda 2009). We find that there is no magic bullet – a combination of tools is likely to be optimal with the exact mix depending on the balance of different frictions. But we also show that some tools are inferior in some circumstances from a public policy perspective. To the best of our knowledge, our economic reasoning is new to the literature, and apparently to some policy makers as well. At the same time, it uses very basic economic and financial principles. Just using a modified Modigliani-Miller framework and assuming that the government aims to restore the systemic banks to a given solvency target, we can differentiate the value proposition of various restructuring tools. For example, unless assets are priced well below their fundamental values, it is easy to show that an asset sales program is likely to be more costly for taxpayers than other forms of restructuring are (e.g., recapitalisation).1
Modigliani-Miller’s value conservation principle is a good start
We recognise, as several economists did before us, the fact that, in principle, restructuring to improve bank’s solvency can be achieved without taxpayer transfers. For example, by negotiating with bondholders to convert some debt into equity, banks’ solvency can be improved. Zingales, for example, has been advocating for mandatory debt-to-equity conversions. The separation of a bank into a “good bank” and a “bad bank” has been discussed on VoxEU by Jeremy Bulow and Paul Klemperer. However, in practice renegotiation can be very hard and lengthy for large systemic banks. Debt renegotiation may be impossible for a large bank due to difficulty coordinating among many stakeholders, the need for speed, and concerns about the systemic impact of rewriting debt contracts. Consequently, taxpayer transfers are usually needed. How the public funds are injected matters a great deal for the distribution of gains and losses.
Absent renegotiation on debt, a restructuring that lowers the probability of default will largely benefit debt holders. As long as the value of bank assets remains the same (i.e., in a Modigliani-Miller world), debt holders’ gain is equal to shareholders’ loss.2 Shareholders would not approve such restructuring without some taxpayer transfers. The required transfers are then equal to debt holders’ gain.
When taxpayer funds come into play, the social benefits should of course outweigh the costs. The primary public policy concern is to avoid a systemic event triggered by a default of a big bank. The social benefits from avoiding such an event, however, have little to do with the recovery rate on assets, which benefits debt holders. At the same time, the cost of a restructuring plan depends on both the default rate of the assets and the recovery rate of assets in default. Among the plans that achieve the same target level of default rate, those plans that raise the recovery rate more are thus more costly to the taxpayers.
Restructuring options may vary greatly in their costs to the taxpayer
Some restructuring schemes increase the recovery rate of debt in case of default too much. The recovery rates associated with various tools can be easily illustrated in a simple world, a frictionless two-period model. Figure 1 depicts the recovery rate as a function of asset payoffs A in the second period for different restructuring schemes. Absent restructuring, the probability of default of the bank is the probability that the asset value A will be less than the debt obligation D in the second period. Assuming that the government’s goal is to decrease the bank’s default probability to a certain target level in the first period, it has to decrease the threshold triggering default from D to A*. Using this same target default rate, the effects of three schemes can be summarised as follows.
Figure 1. Debt recovery rates
- Recapitalisation: New cash is added to the bank’s assets by issuing new equity. The value of the new cash and the value of new equity issue must be equal if it was a market-based transaction – if the market value of the equity issue is $100, then exactly $100 cash is injected in the bank. Therefore, the bondholders and old equity holders are still claiming the same present value of assets as before. The only difference is that, with new cash, default is less likely – the triggering point moves from D to A*. Even with a zero-value realisation of the asset, the cash can be recovered perfectly by debt holders and thus the intercept goes up to D–A*. The recovery slope remains at 1 (middle orange line), the same as before (lower red line). The difference between the red line and the orange line is the gain to debt holders, which is equal to the shareholder loss.
- Asset guarantees with cap: The same recovery schedule as with recapitalisation can be achieved if the government guarantees asset values up to a cap on the maximum transfer from the public sector to the bank of D–A*. For example, the government may guarantee the value of total assets for the first 10 percent loss, but not more.
- Asset sales: The recovery rate is higher if the bank is restructured through asset sales. Suppose a fraction x of risky assets are converted into cash, then the assets as a whole become less risky. Under this scheme, the recovery value will change less than one for one with the value of the bank’s initial assets. Any additional loss of $1 on the original asset decreases the recovery value by only $(1–x), so that the slope of the recovery rate becomes 1–x (upper blue line). Apparently, more value is transferred to debt holders from taxpayers (the area between the upper blue line and the middle orange line) in an asset sales program compared to recapitalisation or asset guarantees. Accordingly, a larger subsidy is necessary and the scheme is less efficient from a taxpayer point of view.
Economic reasoning can help to evaluate the pros and cons of a specific restructuring plan, especially from the viewpoint of taxpayers. A sound analytical framework can help differentiate among several restructuring options from the viewpoint of the taxpayer transfer. This is rarely done, however. At the same time, casual empirical evaluation is often used, yet is very misleading. For example, reactions in bank stock prices cannot be used to judge the success of a restructuring plan – an increase in bank stock price after the announcement of a plan may well signal excessive taxpayer transfers.
Minimising the public transfer is important, not merely because it lowers the fiscal burden but also because it affects the way the future financial system will operate. Besides high direct costs, boosting banks’ shareholder value more than necessary by taxpayer transfers encourages future excessive risk-taking. Another lesson therefore is that to forestall future financial crises, managers, shareholders, and creditors should be held accountable and face punitive consequences.
Bulow, Jeremy and Paul Klemperer (2008). “Reorganising the banks: Focus on the liabilities, not the assets”, VoxEU.org, 21 March.
Hart, Oliver and Luigi Zingales (2008). “Economists Have Abandoned Principle”, Wall Street Journal, December 3.
Landier, Augustin and Ueda, Kenichi (2009). “The Economics of Bank Restructuring: Understanding the Options,” IMF Staff Position Note, SPN/09/12.
Pomerleano, Michael (2009). “The deleveraging process is inevitable”, VoxEU.org Global Crisis Debates, 6 July.
1 Risky assets may be underpriced vis-à-vis their fundamental value when investors face shortage of funding (“limits of arbitrage”, see ICMB-CEPR Geneva Report, summarised on VoxEU by Charles Wyplosz.) In this case, the government can profitably buy those assets and hold them long term. Another market failure occurs when no investor is sure about the true quality of assets (the “lemons problem”). If a bank sells toxic assets to the government using a fair-price finding mechanism, markets will become less suspicious about the asset quality of banks. Our results are thus more applicable after market prices stop free-falling and after independent audits or stress tests reveal more of the true asset quality of banks.
2 The same analysis goes through for a more general case in which the value of bank assets increases after restructuring.