by Professor Bart Lambrecht, Director of the Cambridge Centre for Finance and the Cambridge Endowment for Research in Finance
The 2008-2010 financial crisis and the COVID-19 (coronavirus) pandemic have highlighted the importance of an orderly insolvency resolution mechanism (hereafter IRM). Liquidation and bailout were the main IRMs prior to the crisis. In response to the criticisms on the bank bailouts during the financial crisis, regulators on both sides of the Atlantic devised new regulatory frameworks that attempt to minimise the use of public money to recapitalise failing banks. In this context the bail-in tool is probably the most important regulatory innovation. Bail-in is a statutory power in the hands of resolution authorities that permits them to write down part of the bank’s liabilities or to convert the bank’s liabilities into equity in order to preserve the bank as a going concern.
Existing papers on IRMs typically focus on one single type of IRM (usually liquidation or bailout). We lack papers that compare how firms behave and fare differently under the three IRMs. Important questions remain therefore unanswered. For example, conventional wisdom might predict that the guarantees and safety net provided by bailouts encourage excessive lending and risk-taking compared to liquidations and bail-ins. Is this indeed the case? How do different IRMs affect payout, leverage, the loss in default, insolvency rates and the net value created by a bank? A comparative analysis of this type is missing. Prior studies have also ignored the role of IRMs for payout despite the fact that regulatory restrictions on bank payout during the financial and COVID-19 crises highlight the importance of payout policy for stress testing.
Lambrecht and Tse (2021) develop a unifying, dynamic model for IRMs that addresses the above questions from a micro-prudential perspective, ie they study the effects of insolvency regulation on an individual bank. The paper explores how the three IRMs affect the payout rate, as well as the quantity and quality of loans when these three decision variables are set by risk averse inside equity holders. Insiders can invest in risky assets (loans) of which the return follows a jump diffusion process. The diffusion component reflects continuous shocks to loan returns, whereas the jumps correspond to rare, negative shocks (hereafter referred to as “crashes”). Crashes arrive according to an exogenous Poisson process, but the fraction of the assets that is destroyed by the crash (ie the crash risk exposure, a proxy for “loan quality”) is a decision variable under insiders’ control (eg through collateral requirements). Assets with a higher exposure to crash risk carry a higher expected return.
The paper’s model shows that, from a micro-prudential perspective, banks adopt the lowest payout rate and create the most value net of any recapitalisation costs under the bailout regime. On the downside, insiders have a stronger incentive to put banks at risk of insolvency, increasing the insolvency rate within the banking sector. The exposure of bank assets to crashes can, however, be kept low by giving insiders skin in the game in the event of a bailout. Excessive risk taking can be curbed by penalising (rather than rewarding) managers for failure. To avoid bailouts with public money, a fraction of total bank payouts during good times can be put in a bailout fund to cover expected bailout costs. This can be implemented through a proportional tax on payouts, without distorting insiders’ incentives. Such a bailout fund is viable if banking (net of recapitalisation costs) is a positive net present value (NPV) activity before and after the bailout. Furthermore, a bailout fund could resolve bank insolvency in a relatively speedy fashion.
Under the liquidation regime banks are least prone to insolvency but they incur the largest loss given default. Since managers receive nothing in liquidation and enjoy limited liability, they do not care about the size of the loss in default. A regulatory regime shift from bailout to liquidation therefore not only reduces the insolvency rate within the banking sector, but also increases the loss given default. This highlights a regulatory tradeoff that hitherto has not been recognised.
If the aim is to keep the amount of lending as well as the banks’ exposure to crashes low, then bail-ins with debt-to-equity conversion can be a superior alternative to liquidation or bailouts. The price to pay is that banks grow more slowly and generate less value under this bail-in regime. Lambrecht and Tse (2021) show that bail-ins with straight debt write-down create incentives for higher leverage, higher crash risk exposure, higher cost of debt, higher payout, and lower insider claim values than their debt-to-equity conversion counterpart. The increasing usage of principal debt write-down (PWD) bonds in recent years might arguably be a cause for concern with regulators.
The model also highlights a number of caveats associated with bail-ins more generally. First, banks need a sufficient amount of unsecured creditors that can be bailed in to avoid a bank run. The bail-in unravels if depositors are at risk. Second, whether bail-ins mitigate managers’ incentives to issue low quality loans, depends on managers’ payoff in a bail-in. As with bailouts, it is important that managers’ fortunes remain closely linked at all times to the state of the bank; managers may have to be punished in the event of heavy losses. The bank recovery and resolution directive (BRRD) stipulates that management should in principle be replaced in a bail-in. The model shows that merely replacing managers exacerbates moral hazard problems if managers have no liability and walk away scot-free. Finally, while a bail-in turns an insolvent bank into a solvent one, it does not inject any new capital (unlike bailouts). Bail-ins may therefore not resolve a bank’s liquidity problems.
References
Lambrecht, B.M. and Tse, A. (2021) “Liquidation, bailout, and bail-in: insolvency resolution mechanisms and bank lending.” Journal of Financial and Quantitative Analysis (forthcoming)