THE world’s banking system is both mindbogglingly complex and too vital to fail. After only a year’s deliberation, the finest minds in governments, regulatory bodies and central banks have decided how to improve the way it is supervised. Their answer, it appears, is thicker insulation and better preparation against folly and accident. In December, under instruction from the G20, the Basel club of bank supervisors published new proposals on capital and liquidity “buffers”. These could be in force by the end of 2012.
The speed of the reaction is impressive and most of the proposals look sensible. Yet a feeling remains that the fine minds have evaded the really difficult question. If the banking system resembles a line of climbers roped together, then regulators are busy making the clothes warmer, the maps more accurate, the rations more filling and the whistles louder. Unfortunately none of that is any good if someone falls over the edge, as a handful of banks are wont to do in financial crises. Unless a way is found to solve this problem, taxpayers will remain destined to rescue the flakiest firms time and time again.
In part the focus on capital and liquidity buffers reflects the poverty of the alternatives. Breaking up banks is hard and of uncertain benefit. Having public-sector bureaucrats run them is as unattractive as leaving the discredited masters of the universe in charge. And despite promises that regulators will be cleverer and central bankers more alert, banks are certain to get into trouble again, as they always have. The way to protect taxpayers, the argument goes, is to compel banks to have buffers thick enough to withstand higher losses and longer freezes in financial markets before they call for state help.
The proposals have already been dubbed “Basel 3”—which tells you regulators have been here twice before. Alas, the record of bank-capital rules is crushingly bad. The Basel regime (European and American banks use either version 1 or 2) represents a monumental, decades-long effort at perfection, with minimum capital requirements carefully calculated from detailed formulae. The answers were precisely wrong. Five days before its bankruptcy Lehman Brothers boasted a “Tier 1” capital ratio of 11%, almost three times the regulatory minimum.
That poses an obvious question for bank supervisors: if they have already tried and failed to make capital rules foolproof, why should they do better this time? They do not just have to worry about rules being too slack. If they overreact to the financial crisis and devise rules that are too strict, they may endanger the recovery. And how can supervisors deal with the basket-case banks, for which no reasonable buffer will be adequate?
In the days when banks (and their customers) could not rely on governments to save them, they carried huge buffers to protect themselves against losses and drops in confidence. In the late 19th century a typical American or British bank had an equity buffer—ie, core capital—equivalent to 15-25% of its assets (see chart 1). As recently as the 1960s British banks held more than a quarter of their assets in low-risk, liquid form, such as cash or government bonds.
Over time governments have supplied more protection against disaster. First came liquidity support by central banks; deposit insurance followed; in the latest crisis governments have given all creditors a blanket implicit guarantee. As a result, banks have been prepared to let their insulation wear thin. Going into the crisis, some Western institutions’ core capital was 3% of their assets or less, and less than a tenth of those assets were liquid. Government support may also have given banks an incentive to grow much bigger, so that most European countries now underwrite banking systems several times larger than their GDPs. As Andrew Haldane of the Bank of England has noted, the world has come a long way since 1360, when a banker in Barcelona was executed in front of his failed firm.
Such extensive government guarantees render redundant the normal laws of companies’ capital structure, which dictate that high leverage and over-reliance on short-term borrowing are a suicidal combination. A bank can operate with almost no equity, safe in the knowledge that it will still be able to borrow and raise deposits cheaply, because creditors know they are guaranteed. Furthermore, if a bank knows the state will always provide liquidity if markets dry up, it has a big incentive to rely on short-term borrowing (which is typically cheaper than long-term funds). It follows that if banks in a state-backed system are to have safety buffers, the state must determine their thickness and quality.
Third time lucky?
Since 1988 global capital rules have been set by the Basel Committee, a club of regulators which relies on national authorities to implement its standards. Basel 2, a souped-up version of the original rules, has been introduced by most European banks in the past two years. America’s big banks are on track to implement it by next year. It involves two stages. The first is defining capital: crudely, the gap between assets and liabilities. The second is comparing this with assets. Since not all assets are the same, the rules adjust them for risk, often using complicated modelling: a government bond is regarded as absolutely safe and so needs no capital behind it, but a risky property loan requires lots. The rules say that Tier 1 capital—supposedly, in the main, common equity and equity-like instruments—must be at least 4% of a bank’s risk-adjusted assets.
However, the definition of Tier 1 capital was far too lax. Many of the equity-like instruments allowed were really debt. In effect, the fine print allowed banks’ common equity, or “core” Tier 1, the purest and most flexible form of capital, to be as little as 2% of risk-adjusted assets. In hindsight, says one regulator, this was “very, very low… unacceptably low”. Furthermore, investors lost confidence in the way assets were adjusted for risk to compute a capital ratio. For instance, dodgy mortgage securities could be held with little capital against them, on the basis that credit-rating agencies had graded them triple-A. Risk-adjusted according to Basel 2, they were judged almost as safe as government bonds.
The new proposals go a long way to remedying these failures. The definition of capital will be much stricter. In essence, only pure equity will be included and that after deducting spurious benefits such as tax assets and including nasties such as short-term losses on securities. According to some estimates, that alone could wipe out much of the equity of several European banks, although the changes are likely to be introduced slowly. José María Roldán, of the Bank of Spain, who chairs the Basel club’s implementation committee, says “the more revolutionary we are”, the greater the need for a “slower transition”.
At the same time, risk-adjustment will become less dependent on firms’ own internal models, be harder on investment banks and encourage banks to cross-examine the credit ratings of their assets. For good measure the Basel club has also proposed a new liquidity regime that would require banks to be able to withstand a 30-day freeze in credit markets and force them to become less reliant on short-term wholesale funding. The tests, says an American official, are tough and have been “informed” by the crisis.
The big question: how much?
None of this really answers the all-important question of how much capital banks need. There is a trade-off between safety and economic growth: a bank that took no risks at all would not be much use in providing credit to companies or individuals. Getting this trade-off right is difficult. A 2009 study for Britain’s Financial Services Authority concluded that because periodic meltdowns do so much damage, banking systems should ideally be better capitalised and less volatile than they were before the crisis. The Basel club plans to do its own impact study over the next six months. “It is incredibly important” to get the trade-off right, says Peter Sands, chief executive of Standard Chartered.
Part of the answer, however, is plain: banks should have sufficient capital to survive a crisis as severe as the one Western financial systems have just suffered. This is the sum of two parts. The first is the minimum below which a typical bank loses the confidence of depositors, other creditors and its counterparties. This is largely a matter of mass psychology: the rule of thumb in the markets is that it has perhaps doubled to about 4% of risk-adjusted assets. This already has semi-official endorsement: both America’s and Britain’s recent “stress tests” used this as the floor.
On top of this, banks need enough capital to avoid breaching the 4% minimum in a market meltdown. Here, the experience of the crisis has already produced some guidance. The results of America’s stress tests suggest that the country’s big banks will, through their underlying losses, have eaten up capital worth about 4% of risk-weighted assets. A recent Bank of England study of banking crises since the late 1980s in Japan, Finland, Norway and Sweden found that the average bank ate up 4.5% of risk-adjusted assets.
Adding these two elements together implies that a typical bank should run with core equity capital of 8-9% of risk-adjusted assets, which would be eaten away to 4% during a crisis. Not surprisingly, most big banks are near this point after a bout of equity-raising. America’s four largest banks have core capital of $400 billion, almost twice as much as a year ago.
The view of most, but not all, regulators is that the absolute level of capital in the system is approaching acceptable levels. They still want to add more bells and whistles. Capital requirements for risky trading operations could rise by as much as three times. In anticipation of this, pure-play investment banks such as Goldman Sachs are running with core capital of more than 10% of risk-adjusted assets. To augment the capital rules, bad-debt provisions are likely to be more forward-looking. And how the new capital range is managed is still up for debate. Central banks, with their renewed desire to avert credit bubbles, are likely to take a keen interest. The Basel proposals include sanctions on firms that are close to the capital floor, preventing them from paying dividends.
Working out whether banks have already pre-empted the proposed liquidity requirements is more difficult. Some banks, such as Belgium’s Dexia and Britain’s HBOS, still rely heavily on state funding. But overall a dramatic shift towards long-term funding has taken place. Three-quarters of the balance-sheets of America’s four biggest banks are now funded by equity, long-term debt or deposits.
Yet for all this, a single, horrible truth exists. Because most big banks are too interconnected to fail, and could be brought down by a counterparty, the system is only as strong as its weakest member. Although the average American bank ate up about 4% of risk-adjusted assets in losses during the crisis, the worst banks consumed far more (see chart 2). Citigroup, HBOS and Belgium’s KBC, for example, lost 6-8% of risk-adjusted assets. At the crazy end of the spectrum, Merrill Lynch, which had lots of dodgy securities that were marked-to-market (so that losses were recognised upfront), lost 19%. It would have needed a core-capital ratio of 23% to avoid falling through the 4% floor. UBS lost 13%, implying that it would have required a ratio of 17%.
In part, cautions Bernard de Longevialle of Standard & Poor’s, this reflects the fact that the outlier banks’ calculations of risk-adjusted assets were out of whack. For example, subprime securities were treated as fairly safe. But nonetheless it is probably true that even with the right denominator, these firms would have needed capital ratios of far above 8-9% at the start of the crisis to avoid failure. The Bank of England’s research of other crises points to similar conclusions: the worst failed bank would have needed a core-capital ratio of 18% to stay above a 4% minimum.
Clearly, regulators could simply raise all banks’ capital to a level that would keep even the outliers from failure. But that would be prohibitively expensive. For America’s four giant banks, raising core capital to 20% of risk-adjusted assets could require them to raise an additional $30 billion-odd of annual income (to provide a return on that extra capital). If pushed through to customers, that might raise the weighted average interest rate they charge by roughly a percentage point, from 6% now. That would hurt economic growth.
An obvious answer to the problem of outliers is to impose losses on the riskiest banks further up banks’ capital structures, so that creditors rather than taxpayers suffer. Most banks already have additional slices of capital above equity. For example, at the end of 2008 Britain’s firms had core capital (tangible common equity) of about £200 billion ($290 billion), and on top of this another £200 billion of “quasi-capital” consisting of such exotica as hybrid capital and “Tier 2” securities. In essence these are junior forms of debt which in a bankruptcy would be hit before senior creditors, depositors and customers.
The trouble with these instruments, however, is that if they end up bearing losses, the entire bank is usually judged to be in default, causing a stampede of counterparties, depositors and other senior creditors who fear they will lose too as the bank is wound up, destabilising the system as a whole. In the jargon, these instruments are unable to bear losses while a bank is a “going concern”.
Ideally, then there would be a layer of creditors who could absorb losses while the bank remained in business. The most concrete idea so far is “contingent convertible” capital, or Coco: debt that converts into equity if a bank gets into trouble. In November Britain’s Lloyds, which took over HBOS and was bailed out by the state in 2008, issued a Coco bond equivalent to 1.6% of risk-adjusted assets. It pays a coupon, like a normal bond, unless the bank’s core capital falls below 5% of risk-adjusted assets. At that point the coupon is cancelled and the bond converts into equity, boosting the bank’s ability to absorb losses while remaining a going concern. According to a London banker, people “all over the City are working” on similar ideas. Goldman Sachs has indicated it would consider issuing Coco bonds.
Cocos sound too good to be true, which is precisely what worries some observers. The idea “has the feeling of being a silver bullet,” says a lukewarm American regulator. Lloyds is paying a fairly high coupon of 10-11% to attract buyers to this novel security. That is almost as expensive as equity. And executives at other banks and some regulators worry that under extreme stress complex instruments rarely work as intended. Triggering the bond itself could cause a run: counterparties could take it as a signal that the bank was in severe trouble. Coco’s defenders tend to dismiss this risk—wrongly. In a crisis the degree of uncertainty about worst-case losses and mistrust of banks’ accounts becomes so high that counterparties run after any admission of trouble.
There are other niggles. The capital ratio at any given moment is highly dependent on when managers write down bad debts. A European bank boss paints a nightmare picture of Coco investors buying insurance “wrappers”, offloading the risk to another counterparty, in much the same way that American International Group, a once-mighty insurer, became a rubbish dump for the “tail risks” no one else wanted. If that happened to Cocos, they might merely shift losses from one place to another—and save taxpayers nothing in the end.
Between going and gone
Cocos deserve a chance, but there is an alternative. This is to try to create a middle state for banks between going concern and gone: a partial bankruptcy. Over the past year there has been much talk about creating “special resolution regimes” and “living wills” for failing banks. Many of these ideas are well-meaning waffle, little better than glorified contingency planning. Also of little use are some more macho notions doing the rounds. Simply giving a resolution authority the right to beat up all creditors would ensure that any bank at risk of falling under its auspices would face a run.
One option is to ring-fence the deposit-taking parts of banks and offer them a full guarantee. This would amount to a stealthy reimposition of the Glass-Steagall act, the Depression-era law that separated American commercial and investment banks, and would be hugely complicated. The alternative is to force banks to issue bonds that would automatically suffer partial losses in the event of state intervention, a little like Cocos. Either way, the objective would be to guarantee enough of an institution’s balance-sheet to avoid a run, while leaving enough of it without a guarantee to protect taxpayers from even the outlier banks.
That is not an easy balance to strike. But the risk now is that regulators retreat into designing cleverer ways to make the average bank safer, while ignoring the greater question. That is not how to make regulation cleverer, but how to protect taxpayers from a huge bill when all the precautions fail and a bank steps into the void.