Courtesy of Andreas Wesemann
Deposit insurance has been the most sacrosanct component of the regulatory regime for banks since it was widely introduced in Europe during the 1970s following the collapse of the Bretton Woods regime. While designed to protect consumers against undercapitalized banks, and their tendency to collapse in times of stress, deposit insurance has also encouraged such undercapitalisation.
Since the early 1970s, the number of countries with deposit insurance has increased tenfold, while the number of banking crises worldwide has risen by a factor of almost 500 compared with the preceding 25 years. There are many factors that have contributed to this increase in financial instability such as the collapse of a nominal anchor via a fixed exchange rate system, financial deregulation stimulating rapid credit growth within and across borders, and, I would argue, misconceived financial regulation which substantially increased the scope for, and aggravated the consequences of moral hazard and the mispricing of risk.
Deposit insurance has been a typical example of this. In the UK it was introduced in 1979 via a European Community directive (although the UK government had previously considered it for a number of years). This was in response to several banking crises in Europe during the 1970s, including the secondary banking crisis in England in the early 1970s, the collapse of Michele Sindona’s banking empire in Italy and the US in 1974 (which was triggered by the insolvency of Sindona’s Franklin National Bank in Long Island), and a Spanish banking crisis that commenced in 1977. It is noteworthy that in none of these cases, did retail depositors actually lose any money – in fact, at least in the UK, there is no precedent ever for a bank collapse where retail investors lost deposits. The failure of City of Glasgow Bank (CGB) in 1878 was the last banking crash that caused a “run” (if you exclude the very short run on Northern Rock in 2007) and depositor losses, but the deposits in question were those of some of the leading shareholders in CGB who had been responsible for its demise.
The UK history tells a more general story: real industry-wide “bank runs” are extremely rare. In reality, banking panics have usually been associated with depositors moving their money from unhealthy banks, and those associated with them, to healthy ones. The bank runs in the US in February and March of 1933, in response to the imposition of bank holidays, are probably the closest to a general bank run that can be found in the history of the US. All other bank runs in the US during the depression were restricted to money being moved by depositors from bad to good banks. This suggests that the problem of contagion – where healthy banks are brought to their heels alongside bad ones as panicking depositors withdraw their money from all institutions indiscriminately – is much less prominent, and much less of a risk, than commonly perceived. This undermines one of the foundational arguments in favour of deposit insurance. As Calomiris and Mason put it:
Deposit insurance and government assistance to banks since the Depression have been motivated in part by the perception that bank failures during the Depression were a consequence of contagion, rather than the insolvency of individual banks. If private interbank cooperation, buttressed by liquidity assistance from the monetary authority […] is adequate to preserve systemic stability, then a far less ambitious federal safety net might be desirable.
Today the federal safety net is, by contrast, extremely ambitious. For instance, the large majority of deposits in the UK and the EU are fully insured. The European Commission has estimated that 90% of all deposit balances in the EU are lower than its insurance threshold, and calculations from the British Bankers’ Association in 2008 suggest that almost all retail deposits in the UK are covered by this limit. In other words, the vast majority of individual account holders are not exposed to any theoretical loss in the event of a bank insolvency.
There can be little doubt that the existence of almost complete loss protection for the vast majority of depositors affects their behaviour and willingness to undertake risk assessments. Furthermore, it is completely natural to expect that depositors will translate a statutory promise of partial compensation, in times of crisis, into a pledge of unlimited protection during real duress. This is exactly what was offered by politicians in many countries during the 2007-09 crisis and has happened in other banking crises before. On the basis that this shifts the economic cost of crises without reducing it, such depositor protection is equivalent to a substantive mispricing of the risks associated with depositing money with banks – in other words, the macroeconomic equilibrium cost of retail deposits is higher than prevailing market rates. A secondary consequence is that the volume of such deposits placed with banks is almost certainly higher than it would be in the absence of deposit insurance. A nonchalant attitude towards risk by depositors encourages a greater degree of nonchalance by other creditors and banks themselves. It is natural for banks to seek to maximise cheap leverage of this kind (witness, for instance, the loan/deposit ratios far above 100% prior to the crisis). Banks can very precisely control, via management of their position on “Best Buy tables” for savings products, the amount of retail funding they raise and in effect, the availability of such funding is almost unlimited. Deposit insurance therefore subsidises an already cheap form of debt whose supply is almost unlimited for banks (the Bank of England estimates that this subsidy amounted to £30-50bn p.a. in 2010). Therefore, banks have historically been able to operate at very high levels of leverage with limited constraints on funding balance sheet growth.
The problem of moral hazard is aggravated by, and aggravates, the difficulty of correctly pricing risk and dealing with uncertainty. Tversky and Kahneman’s “availability heuristic” (where probabilities are affected by ease of recall and mental association) and Herbert Simon’s “threshold heuristic” (low levels of probability are treated as being zero) appear to be central elements of human decision-making and risk-assessment processes that lead individuals to systematically underestimate “fat tail risks”. These difficulties of complete risk assessment highlight Robert Merton’s 1977 theoretical confirmation that banks will hold riskier assets when deposits are insured. One does not even have to believe that moral hazard is a direct function of deposit insurance. It is sufficient to argue that it is a reflection of the significant leverage in banks – ultimately attributable to deposit insurance – that generates moral hazard. Adding extensive insurance protection to disaster myopia and leverage scales up risk exposures (i.e. probability of defaults) and potential losses (i.e. loss given defaults) significantly. This has been known for a long time.
Common sense and sophisticated “theories” suggest that deposit insurance is a bad idea – and that it has contributed to the frequency and severity of banking crises since the 1970s. The evidence fully supports this common sense-cum-theory prediction. Moments in history where deposit insurance was not compulsory allows for a comparison of the performance of insured banks against a control group of uninsured competitors. This is possible in relation to some of the voluntary insurance schemes that existed in the US until the mid-1920s. David Wheelock and Paul Wilson have analysed the characteristics of banks in Kansas that participated in an optional insurance scheme from 1909 until 1929 in the wake of a decade-long banking crisis, and compared it to uninsured Kansas-based banks. The results are illuminating. Between 1920 and 1926, 122 state-chartered banks failed in Kansas, of which 94 had been members of the insurance scheme (representing a failure rate of 4.6%) and 28 had not (a 2.3% failure rate). By comparison, the failure rate of the uninsured national banks in Kansas was only 0.8%. The proportionally higher failure rate for insured banks reflects the greater riskiness of their operations: insured banks had lower capital ratios than uninsured banks, higher deposits relative to assets, and held fewer liquid assets against deposits than uninsured banks. As the authors conclude: “Conservatively managed banks were less likely to fail and, at the same time, banks that carried deposit insurance were more risky and, hence, more likely to fail than their uninsured competitors.”
The Kansas experience from almost a hundred years ago is typical of many insurance schemes. Asli Demirgüҫ and Enrica Detragiache, two leading specialists on deposit insurance, concluded, based on data from a large panel of countries from 1980 to 1997, that “explicit deposit insurance tends to be detrimental to bank stability…the more extensive […] the coverage offered to depositors, where the scheme is funded, and where the scheme is run by the government rather than by the private sector.” They also found that insurance for one creditor encourages other wholesale creditors to pressure policy-makers to extend protection to their own claims. Building on earlier work by one of the authors in the late 1990s, which found that banks’ costs of funds were lower and less sensitive to bank-specific risk factors in countries with explicit deposit insurance, they concluded “[…] that deposit insurance weakens market discipline, be it discipline exercised by depositors, by other bank creditors, or by bank shareholders.”
In summary, common sense, theory, and real world evidence suggests that deposit insurance shouldn’t exist.
A Reform Proposal
A simple and effective way of eliminating deposit insurance would proceed with five elements:
- An announcement of the abolition of deposit insurance (“ADI”) after two years. Pre-announcing it in this way gives everyone enough time to prepare. A gradual phase in approach, over a number of years, has no clear benefits (market participants basically operate as if a policy has to be implemented on day 1 even if the actual implementation date is far in the future).
- Concurrently, legislation is passed that the makes the payment of any compensation to depositors for losses they may suffer in the wake of a bank insolvency illegal. Unlike implicit deposit insurance in countries like Israel, that do not have an explicit deposit insurance scheme, where depositors can reasonably assume that the government will protect them in times of crisis, the proposed statute would prevent future governments from doing so. Enshrining the unavailability of sovereign insurance protection in statute ensures that ADI is dynamically consistent, and has appropriate credibility to confirm to depositors, bank shareholders, creditors and executives that the government means what it says and the policy is therefore credible over time.
- Thirdly, new legislation is passed to make all customer deposits (retail and corporate, not including inter-bank deposits) preferential liabilities in insolvency. As the vast majority of deposits are already below the deposit insurance limit, this would not increase this senior funding source materially as a share of banks’ total funding. Depositor super-seniority would mean that losses suffered by banks in a future banking crisis would have to be extremely large before depositors are affected – for the UK they would have to be almost double the largest loss suffered by a major bank during the 2007-09 financial crisis (Anglo-Irish Bank). There is no precedent for losses of such magnitude, and it is not the remit of regulators to plan for, or protect against, losses of such magnitude in any event: risk cannot and should not be eliminated, it should be managed.
- A state-owned savings bank – in the UK this would be National Savings and Investments (NS&I) – begins to offer savings and current accounts to everyone. By establishing – or, as the case may be, making use of – an existing and popular state-backed financial institution (NS&I is very popular in the UK) for common banking purposes the government could create an approximate floor for the capital base a private-sector bank would have to hold to remain competitive. This will turn out to be simpler and more effective than issuing directives or expressing opinions about what “the right” capital ratio for banks is.
- Finally, all deposit-taking banks regularly publish their capital and leverage ratios. These announcements should encourage depositors to consider carefully whether to deposit their money with institutions that have a “low” capital base in comparison to NS&I’s capital ratio.
At the same time ADI is implemented, the Basel III capital adequacy and liquidity regime is also abolished. Capital and liquidity regulation is the price banks pay for government insurance of their largest creditor base; once such insurance disappears, the logic for a sovereign capital adequacy regime evaporates.
Many will argue that the abolition of national bank capital and liquidity regulation is completely reckless: doesn’t ADI make regulation more, not less important? This overlooks its essential economic purpose: the pricing and allocation of risk in the financial system needs to be more effective, and prices paid for services provided should reflect long-term default risks. All protagonists in the banking world need to be incentivised to review fundamentally the way they evaluate, measure and price risk, and this applies to depositors as well as other bank counterparties and shareholders. The history of banking crises in the last few decades suggests that capital requirements under Basel III may still be too low. No other policy will be as effective as ADI in forcing market players to review the capital structure of their operations in order to remain profitable, and stable, going concern entities.
In a world without deposit insurance, risk-averse customers would have recourse to a zero-risk banking option offered by NS&I. Depositors will therefore review where they hold their money and leave only that portion of their deposit balances which they believe they can afford to lose (let’s call these “Surplus Deposits”) with institutions where the probability of loss is greater than zero. This does not mean that all the remaining balances – let’s call these “Core Deposits” – will be moved to NS&I (in the case of the UK) or an equivalent risk-free repository elsewhere. But it will mean that banks have to persuade depositors that the probability of loss is in fact close to zero, or if not, the returns offered to depositors reflect the greater risk they are exposed to. The beauty of the risk-free savings bank model is that it eliminates the need for knowledge (about banking) and the problems of asymmetric information (between bankers and bank customers); anyone can address their constraints in an extremely simple and effective way – bank with NS&I.
If the initial consumer decision is about minimising exposure to catastrophic loss, then banks will have to hold significantly higher levels of loss-absorbing capital than they do today – probably in excess of Basel III requirements – in order to be able to compete and retain customers. They will also have to offer far better rates to their depositors which may eat into bank earnings. On the other hand, better capitalisation will reduce banks’ funding costs, while depositor super-seniority will increase wholesale funding rates. In total, banks’ funding costs will go up a bit, but not by much. What is likely to happen is that banks’ business models will change. For instance, banks will become more specialized as opposed to serving all risk-appetites indiscriminately. Some banks could become 21st century successors of the Trustee Savings Banks, which emerged as low-risk savings banks in Britain in the early 19th century and were an important part of the banking system until their final amalgamation by TSB Plc in 1986. Others could cater to greater risk appetites by offering specialized loan products. What is very doubtful – because it would be irrational – is that ADI makes lending that has positive net present value on an unlevered basis unviable. However, the identity of the lender may change.
New competition from a well-capitalised bank, and customer behaviour, will create strong incentives for banks to re-create and re-apply a new (or, indeed, the current) regulatory regime as part of one or more mutual guarantee schemes, whereby member banks insure each other’s deposit liabilities on a joint basis, subject to certain membership rules. Such schemes could provide additional comfort to customers that banks are financially sound, and reduce their funding costs, and possibly, the levels of capital they must hold. Current “orthodoxy” dictates that deposit insurance must be a state-backed arrangement. However, some of the most successful early insurance programmes were not state-backed, but mutual guarantee schemes. Most notably, this was the case in three US states in the mid-19th century: Indiana (1834-64), Ohio (1845-67) and Iowa (1858-65). These were very successful regimes with extensive and thorough coverage, strong supervisory authority, and strong bank incentives to comply with the rules. There is no reason why something like that can’t be replicated today. If it cannot – and customers vote with their feet – so be it.
A world without deposit insurance once existed and can exist today. New Zealand, for instance, has never offered deposit insurance, except during the peak of the financial crisis, because it didn’t like the distortion of incentives it entails. The ADI proposal set out here is effective and elegant. Effective because it would bring about substantial and positive changes in the way banking systems and, more generally, the infrastructure for creating, holding and investing money operate; elegant because it would do so endogenously without prescriptive guidance by regulatory authorities.
The role of the state would be defined by consumers depending on their risk appetite. It may evolve to provide basic payments and ultra-low risk investment services for a fee, depending on demand. This will encourage competition and stability in the banking system. While central banks can retain some qualitative involvement in certain aspects of bank regulation, regulation will ultimately become the responsibility of banks (central banks would retain monetary policy functions).
As technological developments like distributed ledger open new opportunities for a better configuration of the financial system, it is important for bank insolvencies to be possible. Abolishing deposit insurance facilitates insolvency while also making it less likely due to the incentive it creates for lowering the overall level of leverage in the banking system. It is difficult to see how such a regime is not an improvement on what we have today.
 In the US it was introduced in 1934 in the wake of the great depression which finally created the necessary Congressional support.
 Quoted in Randall Kroszner and William Melick, Lessons from the U.S. Experience with Deposit Insurance, in Deposit Insurance around the World, ed by Demirguc-Kunt, Kane and Laeven, MIT Press (2008) http://economics.kenyon.edu/melick/Research/KrosznerMelickDraft5.pdf , p.6.
 See Robert Merton (1977), An analytic derivation of the cost of deposit insurance and loan guarantees, Journal of Banking and Finance at http://www.people.hbs.edu/rmerton/analytic%20derivation%20of%20cost%20of%20loan%20guarantees.pdf
 See, for instance, Jack Guttentag and Richard Haring’s influential study “Disaster Myopia in International Banking”, IIF (1986) at https://www.princeton.edu/~ies/IES_Essays/E164.pdf.
 Ibid, p. 23.