14 Feb 2013 11:03am

Too much prudence can be a bad thing

Some argue that much bigger provisions would make banks safer, but John Hughes, partner in KPMG’s financial services practice, and Tony Cates, head of audit at KPMG, argue they could hide problems and damage the economy

On 29 November the Daily Telegraph reported that "At last the Bank [of England] is tackling the regulatory fiasco of Britain’s accounting system. Sir Mervyn King has just dropped a bombshell on British banks by demanding a £35bn capital raising. But the bigger shock is the reason why: bank accounts are “dishonest” because Britain's accounting rules are faulty’. The article referred to the November Financial Stability Report of the Financial Policy Committee (FPC), in which the FPC urged the FSA to ensure that banks’ capital reflects a ‘proper valuation of their assets".

The difference between expected losses and provisions runs into many billions of pounds

It is not clear from the FPC’s report how overvalued it believes current balance sheets to be. Much of the language points only to the possibility that future losses may be more than current provisions and looks forward to the expected loss regime being developed by accounting standard setters as a mitigant to this. This regime would allow more scope for future losses to be accounted for, but would continue to require provisions to be evidence based. We have little difficulty with this.

However, other parts of the report suggest that the FPC believes that the difference between expected losses and provisions runs into many billions of pounds. For example they suggest that a simple expected loss calculation would have resulted in £50bn more in provisions at the outset of the crisis. This compares to £26bn of provisions in 2007 and £67bn in 2012. The fact that their core recommendation is that a review of valuations is necessary also implies significant concern.

You might think that, as conservative accountants, we would prefer this second "super prudent" approach. In fact, as responsible business people, we are troubled by it. Booking large additional provisions now would imply that the policy of supporting the economy through low interest rates has not worked. Of course it may not, but this is only one possible outcome and it has yet to happen. Avoiding this outcome will require objective, evidence based information that informs a transparent debate about the risks we face.

Properly compiled, accounts can help here, both by clarifying the present and by identifying future risks. Stuffing them with provisions to cover uncertain and unquantifiable outcomes will have the opposite effect: accounts based on opaque, unreliable forecasts will mislead and probably make things worse. Accounts need to be used in a very different way if they are to inform good policy making.

Something new

It is necessary at the outset to dispel the myth that UK GAAP allowed much more prudent accounting than IFRS does. This is nonsense. UK GAAP and IFRS are almost identical in the areas that concern commentators. The argument for super prudence is something very new: a clear bias towards a pessimistic assessment of future outcomes. This needs to be justified.

Readers should also be clear that increased provisions do not make banks safer. Only capital can do this, because only capital absorbs losses. Capital requirements are determined independently of accounting provisions using models approved by the regulators. If the FPC wishes to raise capital buffers, it can do so, irrespective of what accountants do. Indeed one of the problems faced by the financial system is that, five years on from the crisis, regulators have still to finalise what they consider appropriate levels of capital to be or to offer much empirical evidence to support such a judgement. The FPC’s report highlights the low share prices of banks and this is a key factor explaining them.

Evidence not speculation is required

 While accounts cannot ensure that a bank can absorb all future losses, they can illuminate its financial health. Provisions are a key part of this, but care needs to be taken in their use as acquisitions accounting in the 1980s illustrates. Provisions were set up on acquisition to cover the costs of putting the acquirer and acquire together. Often these were greater than the acquisition costs justified and the surplus was released to mask business underperformance. By the time this was discovered it was usually too late to do anything about it. This was a concern of both UK and international standard setters, when writing the current accounting rules for provisions. As the ASB said, "The effect of such 'big bath ' provisions has been not only to report excessive liabilities at the outset, but also to boost profitability during the subsequent years, when the liabilities are in fact being incurred."

There is a real risk that super prudent accounting for banks would fall into a similar trap. Such provisions may not be deliberately manipulated, but they would always be unreliable, because they would be trying to predict an uncertain future. This could be potentially very damaging. If they proved too much, a bank’s performance would be understated, reducing its ability to raise capital or to lend and depressing its share price. If they proved too little, they would mask rising loan losses and a bank’s vulnerability until they were exhausted, thereby potentially delaying mitigating action to a later stage, when more damage had been done. A cautionary tale of how predictions of the future are no guarantee of safety lies in the praise given early in the crisis to Spain for incorporating counter-cyclical buffers into its banks’ capital calculations. As we know these proved inadequate and the Spanish banking system had to be recapitalised.

The further you try to peer into the future the less reliable your estimates become

There is therefore no justification for large provisions unless they are based on reliable information. Unfortunately we can find very little reliable evidence for large ‘super prudent’ loan loss provisions in the FPC’s report.

Its summary of the global financial environment and the risks to the financial system provides context. In summary it suggests that government policy has succeeded in nursing the economy through troubled times to date, but conditions remain very fragile. The policy may fail, but it may also succeed, and currently we simply don’t know what will happen. As the FPC says in respect of the eurozone, "it is impossible to determine in advance exactly how risks may crystallise or the precise impact they would have on the UK banking system".

The more explicit evidence is no more convincing. For example the assumption that historical losses incurred between 1990 and 2010 are a reasonable predictor of losses today is hardly reliable because conditions today bear little resemblance either to the early 1990s recession or to the very benign period that followed.

Why then should accounts report, as if it were evidence based, that significant losses will occur? If users assume this reflects reality, they could be frightened into withdrawing investment and killing a nascent recovery. If they assume it is merely a subjective forecast, they are left none the wiser as to what is actually going on or what the alternative outcomes might be.

Rather than trying to predict the future, we need to focus on risk and how we mitigate it. The fact is that UK bankruptcy and insolvency trends today are stable or falling and liquidity in most capital markets is reasonably good. These facts are a consequence of the current low interest rate environment instigated by policy makers, and the starting point for any debate about the risks that it may entail.

Surely the accounts should reflect this to help stimulate that crucial debate, not prejudge its outcome using unreliable numbers?

The limitations of expected loss

 Although the ‘expected loss’ provisioning model has merit, it is not able to predict the future and provide the certainty of outcome that everyone craves. Rather it is a more small scale, practical improvement on today’s incurred loss approach. Today for example, under the incurred loss model we cannot provide for the potential effects of an imminent recession, no matter how likely it is. This makes little sense. But there is a great difference between providing for such likely future events and for ones that are just possible. The further you try to peer into the future the less reliable your estimates become.

Once they are based on speculative forecast, they will almost certainly be unreliable and have little informational use. Most economic forecasts, including the Bank of England’s, are expressed as a range not a point number as a consequence and outcomes not infrequently fall outside the ranges even so. In addition any forecast is limited by our own expectations, which may prove false.

For example, no form of expected loss loan model would have protected us from the financial crisis, because the crisis was not expected.

There is a better way to use the accounts

Accounts cannot protect the system by predicting the future: nothing can. But they can shed light on the current condition of the banks and the risks that they run. This helps users to analyse what might happen and inform mitigating policy action. Here is how it could work.

Accounts cannot protect the system by predicting the future: nothing acn. But they can shed light on the current condition of the banks and the risks that they run


The starting point in assessing a bank’s prospects is a clear understanding of conditions today and of its current trajectory. Without it, any prognosis will be unreliable. Providing this basic, but fundamental, understanding is the accounts’ primary purpose and needs to be maintained. The accounts will make assumptions about the future to account for incomplete transactions, but these need to be grounded in evidence.

The balance sheet, P&L and cashflow are not good vehicles for trying to assess the future, simply because they are not a forecast. Other information in the accounts can help, however, by highlighting risks and much good work is already being done in this area. For example the FSB’s Enhanced Disclosure Task Force recently issued a very good report improving disclosure and the FRC’s new combined code and the Sharman report do much to enhance narrative reporting and interpretation. These initiatives should be built on.

Promoting better interpretation of this data is then key. An important lesson from the crisis is that for some years beforehand the accounts revealed much about the weaknesses in banks’ business models (their reliance on wholesale funding, their exposure to property, the size of their derivative positions and so on). The problem was that few reacted to it. Yet it is critical for the safety of the system that as many people as possible examine the data to maximise the chance that key risks are identified. In particular it is important to focus policy makers on avoiding downside risks early enough for damage to be avoided. In practice this would have had to have happened some years before 2006/7, a difficult task when a booming economy was obscuring increasing risk. Changes to disclosure and to narrative reporting can assist, but they cannot do so on their own and help is needed from other bodies, such as the FPC, to foster this debate.

Auditing too needs to change. Before the crisis, the emphasis was upon checking increasing amounts of data on the assumption that markets would absorb it and price appropriately as a result. In fact pricing did not reflect all the risks in the data and auditors need to spend more time interpreting, rather than just checking the numbers, to help users understand them and to improve our ability to identify threats to a company’s survival. This does not mean that we will anticipate the next crisis, but we can do more to highlight warning signs.

A reality check

The future is more than usually uncertain. We have to acknowledge that accountants are not soothsayers, who can magically forecast it with an accuracy that eludes everyone else. Achieving the right balance between the safety of our financial system and the need to stimulate growth is extremely difficult and requires a careful and transparent assessment of future risk.

Accounts can only contribute to this by reporting objectively and transparently on the evidence available and then prompting a thorough debate, that distinguishes clearly between fact and speculation. Pretending that large provisions against unquantifiable future risks will protect us is the opposite of this.

It amounts to a form of Russian roulette that is potentially very dangerous. It is not just bad accounting, it is positively imprudent.


John Hughes, partner in KPMG’s financial services practice, and Tony Cates, head of audit at KPMG