The trouble with deposit insurance
- Executive summary
- Insurance versus wager
- 'Deposit insurance' is not insurance
- Risk-based pricing
- Further reading
- About the author
- Deposit insurance has been proposed as a remedy for losses arising from bank failures.
- Insurance was invented in the fourteenth century by Italian merchants. Its structure has remained relatively unchanged ever since.
- Insurance reimburses money actually lost due to specific events, but only after payment of a premium based on a probabilistic calculation of the likelihood of loss.
- If no such probabilistic assessment of risk can be made, true insurance cannot be underwritten, although wagers can be made upon uncertain possible outcomes.
- Deposit insurance, which protects depositors' bank balances, is not genuine insurance, because no probabilistic assessment of risk can be made.
- Existing organisations which provide deposit insurance have insufficient assets to indemnify lost deposits in cases of multiple bank failures, and in practice rely instead upon state guarantees.
- The private insurance industry has assets grossly insufficient to insure bank deposits.
- A risk-based assessment of premiums for deposit insurance, if it was to provide adequate income for the insurer, would make deposit insurance prohibitively expensive.
- The state, therefore, is the only body in a credible position to guarantee private bank deposits.
Banks sometimes fail, a fact brought home dramatically during the latest financial crisis. In 2008 and 2009, some 308 US banks failed. Between them they held deposits equal to seventeen percent of US GDP, and included five of the fourteen largest US deposit-holders. Happily the individuals who had entrusted their cash to these banks for safekeeping did not lose a cent of it, since government interventions rescued their deposits. Perhaps policymakers were primed: bank failures are relatively familiar in the US. The savings and loan crisis of the 1980s saw the US government save the day when nearly 4,000 small banks faced insolvency. Depositors did not lose money, despite the widespread financial distress.
The UK experience has been somewhat different. Until the collapse of Northern Rock in 2007, bank failures here were a phenomenon of more distant memory. The Bank of England rescued a number of smaller banks and finance houses during the secondary banking crisis of 1973-5 (while allowing a handful to fail), but one has to look back to 1878 to discover the last instance of the collapse of a major UK bank, when the City of Glasgow Bank folded. During the latest crisis, only four deposit-taking institutions based in the UK have required intervention, but although the number is small, two of the banks were huge. Royal Bank of Scotland and Lloyd's TSB were, by assets, ranked first and fourth in the country. Between them, before the crisis, they held 26% of the UK savings-account market. The banks were not allowed to fail: the government stepped in, and as in the US, depositors' funds were not lost.
How is it that private deposits did not evaporate during these great financial calamities? Part of the answer, perhaps, lies in the safety-net known as deposit insurance, invented in 1829 by the US state of New York, and modelled loosely on the regulated mutual guarantee system imposed on the hong merchants of Canton. These schemes covered deposits, as well as banks' circulating notes. In principle, if depositors know that their money is safe and guaranteed, they are less likely to cause a 'bank run' by rushing to withdraw their cash when an individual institution's fiscal strength is called into question. After thousands of US bank failures during the Great Depression, the Federal Deposit Insurance Corporation (FDIC) was introduced to prevent runs on banks. The FDIC guarantees private deposits of up to $250,000 per depositor per insured bank. The UK also has a deposit insurer, the Financial Services Compensation Scheme (FSCS). It operates somewhat differently, and covers up to £85,000 per depositor per institution.
Despite the presence of these institutions, during the latest crisis governments in both countries did not rely on them to prevent the loss of depositors' cash. To avoid a wider banking-system failure, they intervened by facilitating mergers of troubled banks with stronger competitors, and by providing liquidity though huge loans and direct cash injections, sometimes becoming major equity holders in the process. The challenge of saving the banking system was not one to be met by the deposit insurers. Their job is not to save banks, but to stop bank runs.
Some voices have objected to the state bail-out of the banks. This paper does not propose to rehearse or debate the arguments in favour of, or opposed to, state involvement in banking systems, or to review the mechanics or merits of the interventions that have occurred. Instead, it considers one idea, proposed as an alternative to state intervention in failing banks: the insurance of bank deposits under conventional, commercial, risk-based insurance principles, 'whereby', as authors Michie and Mollan state, 'banks would have to pay premiums according to the risks they run'. The authors find this alternative more palatable than using public funds to sustain the institutions, and propose that it is time to devise 'insurance for bank deposits that would extricate the UK government from the moral hazard dilemma it has created for itself through the rescue of banks and bank depositors'. But is their proposed alternative feasible? Could insurance protect individual depositors against losses due to bank failures in times of financial crisis? The question is an important one as 'Europe-wide deposit insurance' has been mooted by key figures such as German Chancellor Angela Merkel and European Central Bank governing council member Panicos Demetriades as a potential backstop to inject new credibility into the beleaguered European currency.
Insurance versus wager
Insurance is an old and superbly flexible instrument. In exchange for an advance payment - the premium, almost always a percentage of the total amount insured - it provides contingent capital in case of actual loss. Commercial insurances allow businesses to convert risk into certainty by exchanging a foreseeable, future, extraordinary or catastrophic loss for a present, fixed, cash payment. This contingent indemnity allows businesses to trade with less capital than they need in relation to the risks which they face. Similarly, consumer insurances allow individuals to take risks which they otherwise could not afford, or which could lead to financial ruin. In both spheres, over many centuries, insurers have altered and adapted insurance contracts, called policies, to meet the changing needs of largely risk-averse societies. Despite these adjustments, however, the basic structure of insurance has remained unchanged for about half a millennium. Insurance works.
Very few businesses or individuals in the developed world operate without some kind of insurance, but despite its flexibility and widespread take-up, not all potential losses are insurable. First, the likelihood of loss must be measureable as a risk, rather than simply be known as an uncertainty. The distinction, outlined by the economist Frank Knight nearly a century ago, is subtle but critical. In essence, if the probability of occurrence of a foreseeable loss-event can be determined, it is characterised as risk. If probabilistic techniques cannot be applied sufficiently accurately to the potential event, it must be characterised as uncertainty. If uncertainty is too great, insurance cannot be offered. To be rid of uncertainty, the future has to come to pass in the same way as the past has been (an assumption that economists call the 'ergodic maxim'). Transactions which promise, in exchange for a premium, to pay out a larger sum based on the outcome, but which are characterised by uncertainty, are properly described as wagers. This is true even when they look like insurance in other respects. For example, when the Bristol marine insurance underwriter Abraham Clibborn insured the 'risk' of 'Peace till 14th May 1772', he was simply accepting a wager. Although, in the eighteenth century, such 'wager policies' were frequently underwritten at Lloyd's Coffee-house, they were not insurance.
The financial instruments used today to hedge against future negative outcomes, such as interest rate and stock market derivatives, are also not insurance, even though they are often referred to as such. Instead they should be viewed as offsetting wagers. Under such contracts, the probabilistic understanding of outcomes is insufficient for adequate analysis. If it were sufficient, it would be possible always to anticipate losses, which clearly is not the case. Instead, like the wager policies of eighteenth-century Lloyd's or a bookmaker today laying off her risk, these financial instruments are insurance only in a broad, metaphorical sense. Thus, 'portfolio insurance' - which has been widely condemned for deepening the 1987 financial markets crash - does not provide an indemnity comprising contingent capital. The puts and index futures on which it is based are not insurance contracts, they are offsetting wagers. It was not insurance at all, despite the parallel characteristics between its downside-limitation approach and the indemnity offered by genuine insurance, and notwithstanding the ability to use the terminology of insurance in describing the moving parts in the loss-limitation approach.
A precondition for the creation of an insurance contract is the existence of both willing buyers and sellers. Further, the price must be agreeable to both buyer and seller. When, in 1663, Samuel Pepys went to Change Alley in the City of London to insure an overdue vessel transporting hemp from Archangel in Russia, he 'bid fifteen percent, and nobody will take it under twenty percent'. With only one party to the transaction at the price offered, and no underwriter tempted, the vessel was left uninsured (and its fate unrecorded). Thus, two willing parties must be able to agree a price. Uncertainty may make proposed pricing unacceptable to one of them. If a policy is offered under uncertainty, it becomes a wager policy, and the rate charged is likely to be exorbitant, since the underwriter - unable to charge a fair price based on probabilistic calculation - must instead choose a number with a large and comfortable margin. Such underwriting still occurs at Lloyd's, but cover for uncertainties is expensive. For example, insuring event cancellation due to rain (quaintly known as 'pluvius insurance') usually attracts premiums of 40% or 50%.
'Deposit insurance' is not insurance
'Deposit insurance' is a term applied to guarantees made by government institutions to individuals making private bank deposits. The term is misleading, since the nature of the guarantee does not meet the criteria, described above, of genuine insurance. The chance of bank failure or illiquidity is an uncertainty. Therefore deposit insurance premiums cannot be set in advance by the underwriter based on probabilistic analysis of the likelihood of loss. In Britain, FSCS 'insures' bank deposits made by private individuals and some small businesses to a maximum of £85,000 per depositor, per UK authorised deposit-taker (the limit was increased from £50,000 late in 2010 to satisfy an increased European Union minimum). The compensation payments which the FSCS makes constitute contingent capital, since they replace actual losses. However, the system is not insurance. If it were, the depositors (or their agents, perhaps the banks holding the deposits) would have to pay, in advance, a premium to the FSCS which was calculated based on a probabilistic determination of the probable amount of the total claims arising. This is not possible, because the likelihood of default is an uncertainty.
Instead, the Scheme operates as a claw-back, with neither the depositor nor the defaulting institution responsible for providing the indemnity. The remaining companies subject to the FSCS reimburse the agency for its compensation expenses. In the scheme's own words, 'the amount levied for compensation payments is the amount of compensation paid plus an estimate of the compensation costs we expect to pay in the twelve months following the levy date, assumed to be 1 July each year, allowing for any retained fund balances'. In effect, the financial services industry is insuring the FSCS, by indemnifying its losses, although the industry receives no premium for the potential liabilities it assumes.
The FSCS's funding cap is £4.03 billion (unchanged despite the recent increase in its maximum payout to £85,000). It is informative to compare this amount to the deposits held by a small UK bank which failed during the latest crisis. The UK's Office of Fair Trading states Northern Rock's retail deposits in November 2008 were £17.31 billion, and in December 2007, after a run on the bank, £9.39 billion. Note, as an aside, that the presence of deposit insurance through the FSCS did not prevent the bank run. Perhaps depositors were wiser than they are given credit for: had no rescue been made, and Northern Rock's default on deposits been complete, even the FSCS, self-described as the 'compensation fund of last resort for customers of authorised financial services firms', would have been swamped. The state's implicit guarantee of the FSCS would have been forced to respond. In practice, the FSCS's funding cap represents only a tiny fraction of all of the deposits held by UK banks. According to the Bank of England, total bank deposits of 'individuals and individual trusts' were £989.6 billion at May 2012.
Total net claims made against policies underwritten by Lloyd's of London, the world's largest insurance market, in 2011 were £12.9 billion ($20.6 billion), including £4.6 billion of claims for catastrophes, the largest annual claims total the market has ever faced. It had attracted premiums of £23.5 billion ($36.3 billion), but when reserves for future years' losses, the effect of reinsurance, and investment income are calculated, the outcome was a loss to the market of £516 million. A claim of £4.03 billion, the maximum compensation offered by the FSCS, would nearly match the sum of catastrophe claims in the worst catastrophe year ever for Lloyd's. Such a loss is one that the market could certainly bear, provided it had charged a commensurate premium agreeable to both parties. However, given the uncertainty that characterises the likelihood of such a total loss under the hypothetical deposit-insurance policy, Lloyd's underwriters would have required an enormous rate of premium, just as they do when underwriting pluvius cover.
For deposit insurance to be anything but administrative mechanics hiding actual public indemnity of lost deposit-holders' cash, banks would have to be charged a similar risk-based premium by a public deposit insurer. Michie and Mollan have called for such risk-based premiums to be assessed to support public deposit insurance schemes. Alas, given the relatively small income that banks can earn by holding depositors' cash - on which the banks must pay a return to depositors - it is not clear that such an insurance policy would be economical for the banks. Their earnings on deposits comprise only the difference between the interest they pay depositors, and the income they earn from investing deposits. If banks were to pay the premiums required for unlimited commercial deposit insurance, whether provided under a compulsory government scheme or by the private sector, the UK free-banking model - already under threat - would become woefully uneconomic.
It is not necessary to speculate. The US FDIC has operated a risk-based pricing system since 1993. Individual insured institutions are 'assessed based upon statutory factors that include the balance of insured deposits as well as the degree of risk the institution poses to the insurance fund'. From 2011, annual assessments range from between 2.5 and 45 cents per $100 of deposits. 'Deposit insurance is exactly that: insurance... The higher the risk the higher the premium', an explanatory video on the FDIC web site declares. It is not insurance, however. The rate banks pay into the scheme is not based on probabilistic loss calculations, but on a risk profile calculation, and the FDIC levies funds after losses have been incurred, not in advance, as genuine insurance requires. From the 2010 passage of the Dodd-Frank Wall Street Reform and Consumer Protection Act, the FDIC must maintain a 'minimum designated reserve ratio' of 1.35% of estimated insured deposits.
Thus in 2009 the FDIC charged $10.2 billion in regular assessment revenues, plus an additional $5.6 billion under a 'special assessment', and at year-end required its insured institutions to pre-pay three years' worth of premiums - $46 billion. This is not the way commercial insurance behaves. Further, unlike an insurer, the FDIC is not required to maintain reserves sufficient to cover foreseeable losses. Instead, in a large-loss situation it would be forced to rely on a bailout of its own. It has a $100 billion line of credit with the US Treasury - a form of contingent capital - which it can exercise if its assets fall short of its liabilities to depositors. If that is not enough - as it clearly would not have been, had the five of the fourteen largest US deposit-holding banks which failed been allowed to forego their liabilities, rather than being merged away or rescued by the government in the latest financial crisis - the FDIC ultimately has the backing of the US state. Despite risk-based premium rates, the fees it levies are unsatisfactorily low to provide a true insurance backstop. If they were not, they would almost certainly be unmanageably high from the banks' perspective. The FDIC would have to charge at least its current rates in percentage points, rather than basis points, and do so for several years, to establish a reserve sufficient to cope with the indemnities that would have arisen had the US state not intervened to prevent the collapse of hundreds of banks. When the going gets tough, deposit insurance doesn't cut the mustard, because the total value of private deposits is enormous.
Something similar to the insurance of bank deposits by private financial institutions has been attempted on a large scale - although the experiment failed its first serious stress-test. In the US, a number of conventional insurers (typically known as 'monoline' insurers because they transacted only one sort of insurance) were formed in the 1970s to insure purchasers of municipal bonds against default by the issuer. The insurance served to raise the credit rating of the bonds higher than the rating of their issuers by guaranteeing interest and principal repayments. As the popularity of 'financial guarantee insurance' developed alongside the evolution of the structured finance market, so too did the monoline insurers' appetite to insure a broader range of financial instruments increase. Collateralised Debt Obligations and Mortgaged-Backed Securities were deemed to be insurable by traditional monoline bond insurers, although the nature of the characteristics of possible default events - such as trends in the value of residential property - can be regarded only as uncertainties (at least in retrospect). The 'subprime mortgage crisis' of 2007 triggered significant claims under financial guarantee policies, leading to the collapse of all the leading US monoline insurers. Another major player in the financial guarantee insurance market, AIG, the largest insurer in the world, applied for Chapter 11 bankruptcy protection in 2008, and ultimately survived only because the US government provided a $182.5 billion bail-out package, equal to 1.3% of US GDP.
Thus it appears that providing insurance for depositors against the failure of UK banks is well beyond the fiscal wherewithal or risk appetite of the international insurance sector. Nor do the insurers look like a reliable counterparty in a time of financial-market meltdown. Perhaps other institutions in the financial sector would be willing to wager on the loss of bank deposits, and have sufficient financial muscle to take on a trillion pounds of risk. Such values are not uncommon in some financial markets, and the historical precedent for the sharing of insurance risk among the community of merchants is significant. The banks could insure each other, just as in the fourteenth-century, Italian underwriters of marine insurance policies were merchants who each assumed a share of their fellow merchants' risk. The principle of underwriting was the same at Lloyd's even in the nineteenth century, when underwriters were known as 'merchant-insurers'.
Alas, the insurance of banks by other banks presents an obvious problem of risk aggregation and potential systemic collapse. It was extremely unlikely that a large number of merchant-insurers would suffer serious losses of vessels and cargoes at the same time. When multiple losses did occur simultaneously, such as when French navy and privateering vessels captured the combined Anglo-Dutch Mediterranean trading fleet in 1693, nearly three dozen merchant-insurers in London were thrown into bankruptcy (including the merchant Daniel Foe, who then retreated from active commerce, changed his name to Defoe, and took up writing instead). In globalised banking, it is particularly likely that a domino-effect during a financial crisis will result in 'contagion', with one bank failure following another. The calamities at Lehman Brothers, Merrill Lynch, Bear Sterns, and AIG (technically an insurer, but one which behaved like a merchant bank) prove the case.
Could non-financial companies insure the banks? The combined assets of the five largest companies in the world are roughly equal to the £989.6 billion in private UK bank deposits, so in combination these corporations have sufficient wealth to offer the guarantee. Further, as the business of four of these five mammoth corporations is petroleum, their activities may be sufficiently uncorrelated with financial markets, and demand for their products sufficiently inelastic, to prevent contagion and thus allow their assets to remain intact through a massive financial crisis. However, reimbursing bank deposits would wipe them out. Thus, their directors and shareholders are unlikely to allow the companies to assume such a potentially fatal contingent liability. If they were to do so, the price asked by these corporations would surely be prohibitively high. Private sector counterparties seem in short supply for the insurance of deposit-holders' assets.
'Deposit insurance' institutions which charge a risk-based premium are a nice idea. The pricing approach means their product is more like conventional insurance, the effectiveness of which has been proven over many centuries. Risk-based pricing of such insurance may discourage financial institutions from behaving in risky ways, because the cost of the insurance would exceed the potential profits of deposit-taking. In this way, deposit insurance institutions could behave as a regulatory instrument (although many of the financial instruments which inflicted the most damage on banks during the latest crisis were triple-A rated, and thus would have attracted the lowest risk premium). It is not surprising that some people have seen deposit insurance as part of the solution to future financial crises, and as an alternative to state bail-outs of the banking system in times of strife.
However, it is clear that even when deposit insurance institutions assess premiums based on some measure of riskiness, the amount they charge cannot be based on a probabilistic assessment of default by a deposit-holder. The likelihood of default is an uncertainty which cannot be calculated, as the recent crisis proves: the future does not always follow the same patterns as the past. The collapse of the sub-prime mortgage market was not in anyone's model, for example. The system has too many moving parts, the chance of serial correlations is recognised but unknown, and the behaviour of individuals is not as predictable as many economists would have us believe. Even if the likelihood of default could be determined with accuracy, existing deposit insurance institutions such as the FSCS and the FDIC cannot charge sufficient premium to cover the potentially large losses that could arise in a banking crisis - such as the one experienced from 2007 - which affects a significant proportion of their 'insured' entities. To do so would make the insurance too expensive for banks to buy. The institutions are useful to insure against isolated defaults (although their original purpose was to protect against bank runs, a function which, in the case of Northern Rock, the FSCS failed to perform). However, when a systemic weakness leads to significant failures, an entity larger than the largest deposit insurer must intervene to prop up the system, or the resulting claims will simply swamp the deposit insurers. To prevent this, such institutions usually have a government guarantee, which ensures that deposit-holders will always receive full compensation - but from state coffers.
During the latest banking crisis, governments chose to use their vast assets and unparalleled ability to raise debt to recapitalise or otherwise save the banks themselves, rather than simply to compensate the millions of private losers that wide-scale bank collapses would have created. The entire private insurance sector, and even the corporate sector, do not have an asset base sufficient to perform such a rescue in times of international financial crisis, let alone to insure the deposits held by banks. Further, there seems little incentive for either sector to choose to assume such a threat to their balance sheets, even if the premium offered is commensurate to the most generous estimate of the likelihood of default by banks. There is no non-state counterparty to make deposit insurance work in cases of widespread financial distress, no premium agreeable to both insurer and insured which would yield sums adequate to insure deposits against such a failure, and no satisfactory way to measure probabilistically the chance that banks will default on their deposit obligations. Insurance is not a possible solution to the phenomenon of systemic bank failures. The state is the only institution with sufficient resources to ensure that banks do not collapse, and to 'insure' the deposits they hold.
- A brief history of deposit insurance in the United States (Federal Deposit Insurance Corp., 1998)
- Culp, Christopher, Risk transfer: derivatives in theory and practice (Wiley & Sons, Hoboken, N.J., 2004)
- Elliott, Geoffrey: The mystery of Overend and Gurney: a financial scandal in Victorian London (Methuen, London, 2006)
- House of Commons Treasury Committee, Competition and choice in retail banking, Ninth report of Session 2010-11 (London, The Stationery Office Ltd, April 2011)
- Knight, Frank: Risk, uncertainty, and profit (Houghton Mifflin, Boston, 1921)
- Lybeck, Johan A. A global history of the financial crash of 2007-10, (Cambridge University Press, 2011)
- Raynes, H.E. A history of British insurance, second edition, (Pitman & Sons, London, 1964)
About the author
Adrian B. Leonard is currently an Affiliated Researcher at the Centre for Financial History, University of Cambridge. His work focuses on the financial history of London insurance. Before returning to academia, he spent nearly two decades as a financial writer, journalist, and communications advisor, including for the International Underwriting Association of London. E-mail: firstname.lastname@example.org.
If you have found this paper interesting, why not join our email list to receive occasional announcements when new ones are released?