To the editors:

In 2016, the former governor of the Bank of England, Mervyn King, published The End of Alchemy: Money, Banking, and the Future of the Global Economy, a book presenting some interesting paths for reflection on how to change the institutions that rule the global financial system. Economist Henri Lepage has recently published a first exhaustive critique in the French language of this work, which has been for the most part addressed within the English academic and financial worlds.

While Lepage rightly welcomes King’s realization of the impasse in which the current macroeconomic models of central banks have trapped Western economies, he criticizes King’s approach defending the role of central banks as the final stabilizing institution of a banking system when it encounters a crisis threatening its sustainability. Lepage’s criticism is broadly based and argued, but deserves to be explored further on several points.

On the Failure to Estimate Systemic Risk

Mervyn King wants to change the current role of a central bank from a “lender of last resort” to a “pawnbroker for all seasons,” which would commit itself, in the event of a liquidity crisis, to lend to an illiquid bank enough to cover its withdrawals. In exchange, the central bank would reserve assets held by the affected bank, with a haircut decided by the central bank, based on the estimated volatility of the asset in question. In order for the bank to be assured of help, it would have to undertake not to have more short-term deposits in its liabilities than the value of collateral assets that could be used as securities, after a haircut, by the central bank.

Since the mechanism is not entirely intuitive, it is helpful to illustrate it using a simple example.

If a bank has on its balance sheet the following assets, what should be the composition of its liabilities to enjoy the support of the central bank?

Suppose that, according to King’s system, the central bank estimates that the maximum haircut of asset A, considered low risk, is 10%, and that the haircut for asset B, which is far more volatile, is 70%. The commercial bank would thus be authorized to retain 10 + (50 × 0.9) + (40 × 0.3) = 67 short term deposits. In the case of withdrawals exceeding its liquidity, the commercial bank would obtain, from the central bank, liquidity allowing it to pay its depositors. The central establishment would be paid from the discounted assets when economic times, having become more favorable, allow them to realize a higher value. Unless, of course, the “rescued” bank, having recovered its own health, recovers its assets and reimburses the central bank, minus interest.

To obtain a balance sheet total of 100, if the liabilities of the bank cannot exceed 67 short-term deposits, then it should hold 33 in either equity or long-term loans. In the event that losses degrade its ratio of long resources over short, it should either sell assets, raise new equity, or obtain new long-term loans from other lending institutions or bond markets.

Lepage thinks that this system, which seems at first glance more stable than the current one, would give excessive power to the central bank: the power to decide, for each category of investment, the presumed haircut for those investments. It could then, in practice, choose the economic “winners” and “losers,” and even subsidize certain sectors either by mistaking the estimation of the risk carried, or by yielding to more-or-less-acknowledged external pressures. In addition, interest rates on the loans that commercial banks may grant will depend solely on the central banks’ overall risk estimate of volatility in the economy.¹ They will, in practice, have the power to determine, not only the global envelope of possible investments by banking institutions, but also their nature, sector by sector. King’s proposal therefore presupposes, in order to function, that the central banks have nearly-infallible foresight in determining the cost of risk borne by thousands of different assets. We know from the failures of planned economies that this is impossible. Worse yet, in such a system, the opportunities for error would be numerous.

Suppose that the U.S. central bank temporarily estimates the risks of economic turbulence higher than do its counterparts in Europe, the United Kingdom, Japan or China, and therefore, rightly or wrongly, imposes weaker obligations to hold liquidity on the financial institutions under its control. These institutions will face a higher cost of financing their resources than their competitors, and thus be at a competitive disadvantage. The political and media pressure from financial institutions or politicians on the central bank to be “accommodating” will therefore be high, as it is in the current system, and the world’s central banks will generally tend to underestimate risks.

Moreover, the system proposed by King, far from constituting a great paradigm shift away from the one in force today, would perpetuate the perverse incentives of the Basel ratios. It is worthwhile to recall why financial institutions have thrown out their balances of “structured” mortgage-backed security products, especially those created in response to the regulations in force before 2008. Under the Basel system, financial institutions must maintain equity in their balance sheet in proportion to the risk carried by their assets. This level of risk is assessed by rating agencies. The effects of their biases and conflicts on their estimates have been “discovered” a posteriori.

To pay their investors, financial institutions need to hold securities and investments more profitable than treasury bonds, while avoiding risk ratings that would require them to increase their own equity too much. They therefore created, from scratch, structured credit products that combine debts of variable quality. These are resold to investors in bunches, with interest that varies with the presumed risk. Unfortunately, all of these financial machinations had a cost, which reduced risk margins, and pushed the credit rating agencies, themselves under pressure from clients who issued structured products, to underestimate the risks involved. Thus the entire financial sector found itself short of capital to cover its losses. We know what the result was.

Under King’s system, how can we not believe that the same incentives would not produce the same kind of behavior? King’s system, like the Basel regulations, would force banks, depending on the riskiness of their asset mix, to hold onto resources even longer. The work of financial engineers would therefore consist of issuing “reconditioned” financial products, adapted to this new regulation, whose risk of volatility would be underestimated, compared to the expected return. As was the case before the subprime crisis, it is likely that the major players in the financial sector would structure their assets with this type of product. Regulation would encourage them to do so. This, in turn, would standardize the risk profiles of the banks. If underestimations of these risks in bank balance sheets became the rule, it would carry no less systemic risk than the Basel rules.

Both international competition between central banks and the need to make the perceived risk lower than the actual risk borne by commercial banks could lead to the same disturbances in King’s system of compulsory liquidity reserves as the principle of the Basel ratios. Additionally, a bank manager who is assured of receiving a parachute from the central bank will always be tempted to be less cautious than one who lacks this assurance. As soon as global losses on an asset class, judged to be more or less secure, exceed the haircut established by the central banks, conditions for a new systemic crisis would again be met. King’s proposal does not appear to be a paradigmatic revolution. Instead it is a technical adaptation of a system whose basic philosophical principles would be preserved along with its structural defects.

Should Central Banks Be Responsible for Preventing Bank Failures?

Lepage sees in King’s work an attempt to justify, at all costs, the existence of central banks. They are assigned, without any real justification, the quasi-“natural” role of the bulwark against any risk of disruption in the system financing the world economy.

Is this a natural role for them? I am not discussing here the central banks’ role as monopolistic supplier of money, a role in which they are certainly open to criticism. Instead, consider the role the central banks have appropriated as the “lender of last resort.” Is there another way to guarantee the sustainability of the financial exchanges?

The crisis of 2008 highlighted the “bail-in” rules that allow, in theory, the rescue of imperiled banks without taxpayer intervention and without the risk of seriously disrupting the chain of payments between banks. In such a system, the failed bank first sees its shareholders “wiped out,” then the debts, in order of inverse seniority, are converted into shares. These principles have been implemented both by the Dodd-Frank law in the U.S. and by the directive of the European Union’s “Banking Union.” While the principle appears useful, its implementation leaves something to be desired. The difficulties of the Monte dei Paschi bank might have been an opportunity to implement the banking union directive and to test its real effectiveness. Instead, it resulted in a harrowing rescue plan once again supported by the Italian state, whose own levels of indebtedness should no longer allow for such largesse. Indeed, “bail-in” directives have been badly explained to the public. The subject is, admittedly, a difficult one. Governments are afraid of the reaction of the public if there is risk their deposits will be partially converted into bank shares—even if a bank’s balance sheet would be cleared of long-term debts.

As a result, countries have wisely sought not to involve the taxpayers in rescuing poorly-performing banks. But they have for the most part kept state insurance of bank deposits up to a certain ceiling (€ 100,000 in E.U. nations) and have maintained regulatory obligations on equity as a function of assets held. This is a consequence of the fact that they are still bearing the ultimate risk. The central banks have thus been confirmed in their role as supervisors of commercial banks.

When experts in the functioning of the banking system are questioned, even when they are in favor of the intrinsically-virtuous principle of bail-in, they doubt the effectiveness of its implementation. This is because it is the regulator alone who can initiate it, and the regulator is neither infallible or insensitive to political and media pressure. Nor is it incorruptible.

If this power to declare bankruptcy does not devolve to a regulator whose weaknesses will be too easily circumventable, who then should have it? While counterintuitive, the simplest proposal would be to entrust it to the banking officers themselves, but under a strict regime of personal and civil liability for their own property, and, of course, criminal liability in cases of (duly characterized) intentional mismanagement.

In such a system, the directors (CEOs and CFOs), in return for the high salaries they receive, would be liable for all their personal assets—even if they are only salaried employees of the bank—if they take too long to initiate a bail-in procedure. “Too long,” is when the losses are such that they oblige the state to implement a public guarantee of the deposits of non-professional creditors. Such a guarantee is certainly harmful, but also one that seems, politically, difficult to avoid.² The managers of banks in difficulty would thus have a very strong incentive not to wait for the situation to deteriorate too much, and to file for bankruptcy at a time when implementing the bail-in would affect only the shareholders and professional creditors.

Would this unrestricted personal liability be effective? The Swiss banking system is instructive in this respect. In order to meet the international standards imposed by the U.S. legislature, Swiss banks under the status of “private bank” with unlimited liability, such as Pictet or Lombard-Odier, have had to abandon this status for that of a limited company and publish their accounts. The financial world was thus able to discover that these institutions had particularly high percentages of equity (22% for Pictet in 2014, compared with 8% imposed by international regulators) without impairing their profitability, and without exposing them to the most problematic financial sectors. The bank’s shareholders, who are ultimately responsible for any collapse, have always sought in their management a combination of prudence and performance. Their unlimited liability reassures customers who, in return, accept higher management fees than are charged by other establishments. Empirically speaking, the unlimited liability of the leaders therefore seems an effective bulwark against the excessively adventurous management of commercial banks.

It cannot be ruled out, of course, that an ill-advised officer, unable to recognize a failure or overestimating his or her abilities, might underestimate the level of personal risk and fail to flee in advance or call in the cavalry, hoping that luck or trickery will save the day. Unlimited liability would reduce the risk of bankruptcy being declared too late, but would not eliminate it.

Is It Possible to Dispense with Public Guarantees on Deposits and “Parachutes” from the Central Bank?

Irrespective of the political difficulties involved in promoting it, is it possible to imagine a system where there is no public guarantee of deposits and where no central bank intervenes either as an institution of last resort or as an a priori regulator of banking activity? Would such a system be spared the aforementioned problems?

It seems possible to conceive of a system combining true transparency of banking assets, a bail-in that works, a high level of liability on the part of the officers, and the free choice of the depositor as to the level of exposure to a given risk of bail-in, and all this, without any public intervention other than a court judgment in the case of bankruptcy. The depositor could opt for the total security of his deposits by refusing to include them in the bank’s balance sheet. The bank would then, in this case, only be a supplier of means of payment. Of course, in return, the customer would pay for the service at an appropriate price. The bank would also be strictly required to keep in its assets one euro in cash for one euro deposited.

To lower fees, a customer could accept that the bank will use his deposit. But he would be fully informed regarding the risks for his deposit, no matter how minimal. He could either benefit from the services of free means of payment, or else see his account paid less in return for a slightly higher exposure to the risk of bail-in. Finally, he could, like any investor, lend his money to a bank via paid term deposits or bond purchases, or, taking the ultimate risk, buy shares. For all these resources, the bank would be free to choose its assets without any central agency in charge of evaluation.

The balance sheet of the bank would therefore have a structure that might resemble this following, described here in a simplified manner:

It would therefore be fairly easy to define an “index of protection” for the depositor or investor, equal to the cumulative percentage of those right holders with inferior liability. Such an index would provide a first level of information, even to an applicant who knows nothing of the arcana of bank balance sheets.

Thus, in our example, a holder of a free but unpaid deposit would have an index of protection of 70, corresponding to the 70% of the liabilities held by right holders with lower seniority. In other words, to risk a conversion of one’s deposit into shares, the bank’s assets would have to lose at least 70% of their value. This would imply that all the shareholders and lower-ranking creditors had already lost their shirts. This conversion would allow them to glimpse the possibility of recovering their property (even with a higher value!) if the non-depreciated assets of the bank regained some health. In any case, banking disasters at such a level would be extremely rare.

This index of protection would not be sufficient for more sophisticated investors. In order to enable these investors to more accurately assess the level of risk attached to the banks’ assets, and thus to determine in advance the level of risk they are prepared to bear for the proposed return, it would make sense for investor publications to be required to exhibit greater transparency than is currently the case. This is particularly true in relation to the possession of “structured” products, whose opacity makes it particularly difficult to assess risk. The bank should declare, for each financial product, its breakdown into “primary assets” that generate yield: various debts, stocks, bonds, land, real estate, forests, etc. The role of bank stability assessor would not then devolve to obscure elitist bureaucracies, but would be accessible to thousands of investors. The role of credit rating agencies would change dramatically. No legislation would be required to assign to a small number of agencies the privilege of being an approved credit rater. The valuation would be financed by investors rather than by issuers.

In such a system, the definition of the right ratio between equity and other debts would be left to the shareholders themselves to determine. Obviously, they have a special interest in avoiding any situation where heavy losses might lead to a collapse. No regulator or central bank would be required to establish a level of capital or adequate liquidity. It is those who gamble their own money and who are involved in day-to-day business that are best placed to determine the best ways of doing so, as is the case for any craft. In the absence of regulatory guidance, an additional benefit would be that banks would not be encouraged to adopt the same risk profiles. And, as in nature, a more diverse ecosystem is more likely to be resilient, the errors of a few not spreading to all the others.

Finally, governing shareholders could—in order to increase the confidence they inspire and to reduce payments to lenders—decide what level of loss would involve their unlimited personal liability.

Thus, all participants in the system—shareholders, professional creditors, and depositors—would have something to lose if things went wrong. Such a system, in which all stakeholders “have skin in the game,” but are free to vary their risk profiles, is more likely to be “anti-fragile,” (an expression popularized by Nassim Taleb) than another where central regulators might delay the occurrence of crises, but make them, in fact, much more violent when they do occur.

Vincent Bénard

Henri Lepage replies:

Many thanks to Vincent Bénard for providing us with two essential pieces to complement the analyses in my article.

Bénard is the editor of a blog which, at the time of the 2007/2008 crisis, published a series of articles representing without a doubt the best financial commentary written in France about the events of the time.³ Although trained as an engineer, he has the genius of knowing how to explain in detail and in a precise way the decentralized processes through which the free market economy resolves problems that economists, prisoners of their equilibrium “problématique,” do not generally manage to conceptualize without inevitably needing to appeal to the constraining interventions of the state. In addition to being a member of the Turgot Institute, he is also the author of a remarkable book on the housing economy, where he applies the same approach, one that is sadly neglected by academics. His letter demonstrates that he has lost none of his formidable ability to explain, concretely, the market processes that mathematical models automatically tend to obscure.

In considering King’s proposition to move the central bank towards a more narrow function as a “pawnbroker,” Bénard completes my own criticisms by showing that this paradigm shift would still perpetuate the perverse incentives of the Basel ratios. This would inevitably reproduce the same behaviors that are generally invoked to explain the origins of the financial crisis. It is, in fact, a point that I had never thought about.

But the most important part of his letter is incontestably the second, where he shows how a genuine mechanism of self-regulation of bank stability would work, one that would not rely on a public guarantee of deposits, nor on a parachute role for the central banks. It goes without saying that I fully agree with his project and that I cannot recommend his work highly enough to anyone seriously interested in the banking economy and its regulation.

One small comment, nevertheless, about the subject of the structured credit products that are mentioned in Bénard’s letter, which are usually considered principal instruments that led to the explosion. The problem is not that bankers and financiers have come up with innovative and extremely sophisticated formulas for getting around existing banking regulations, but that the regulation itself has massively—and more—contributed to their spread. I am alluding here to the “recourse rule,” an American amendment from 2001 that altered the Basel rules in such a way as to classify MBS in the category of assets subject to the lowest ratio rules (2%). I have often wondered about the sudden, almost vertiginous, acceleration in the volume of subprimes produced beginning in 2001/2002. The answer is there, in this American regulatory initiative, which is never mentioned—if it is not in the work by Jeffrey Friedman and Wladimir Kraus, Engineering the Financial Crisis and the Failure of Regulation, published in the fall of 2011. It is another excellent example of the regulatory delusion of which Vincent Bénard is one of the best and most constant opponents.

Translated from the French by the editors.